Skip to main content

Full text of "The Art of Illumination"

See other formats


This is a digital copy of a book that was preserved for general ions on library shelves before il was carefully scanned by Google as part of a project 

to make the world's books discoverable online. 

Il has survived long enough for the copyright to expire and the book to enter the public domain. A public domain book is one that was never subject 

to copyright or whose legal copyright term has expired. Whether a book is in the public domain may vary country to country. Public domain books 

are our gateways to the past, representing a wealth of history, culture and knowledge that's often diflicult to discover. 

Marks, notations and other marginalia present in the original volume will appear in this file - a reminder of this book's long journey from the 

publisher to a library and finally to you. 

Usage guidelines 

Google is proud to partner with libraries to digitize public domain materials and make them widely accessible. Public domain books belong to the 
public and we are merely their custodians. Nevertheless, this work is expensive, so in order to keep providing this resource, we have taken steps to 
prevent abuse by commercial parlies, including placing technical restrictions on automated querying. 
We also ask that you: 

+ Make non-commercial use of the plus We designed Google Book Search for use by individuals, and we request that you use these files for 
personal, non-commercial purposes. 

+ Refrain from automated querying Do not send automated queries of any sort to Google's system: If you are conducting research on machine 
translation, optical character recognition or other areas where access to a large amount of text is helpful, please contact us. We encourage the 
use of public domain materials for these purposes and may be able to help. 

+ Maintain attribution The Google "watermark" you see on each file is essential for informing people about this project and helping them find 
additional materials through Google Book Search. Please do not remove it. 

+ Keep it legal Whatever your use, remember that you are responsible for ensuring that what you are doing is legal. Do not assume that just 
because we believe a b<x>k is in the public domain for users in the United States, that the work is also in the public domain for users in other 

countries. Whether a book is still in copyright varies from country to country, and we can't offer guidance on whether any specific use of 
any specific book is allowed. Please do not assume that a book's appearance in Google Book Search means il can be used in any manner 
anywhere in the world. Copyright infringement liability can be quite severe. 

About Google Book Search 

Google's mission is to organize the world's information and to make it universally accessible and useful. Google Book Search helps readers 
discover the world's hooks while helping authors ami publishers reach new audiences. You can search through I lie lull text of this book on I lie web 
at |http : //books . qooqle . com/| 

y. &?•.<. co 


TUntversIt? of Wisconsin 



Saftfll^Ufti'f ^*****4iAilLltltItlIimiJHIHlHtU.HitiiAft*tta,iMIfl4fcia^**^*^ 


Published hy the 


Hill Boole Compctny 

New Yorle: 

Successors to thefSookDeparbnente of tke 

McGraw Publishing Company .Hill Publishing Company 

FVibliahers of Books for 
Electrical World The Engineering' and Mining Journal 

Engineering Record American Machinist 

Electric Railway Journal Coal Age 

Metallurgical and Chemical Engineering Fbwer 

7 titUWU11*********ltlWt11Ul****9i**9********************UHi11t1wm*l 






Fellow, American Academy of Arte and Sciences; M. A. I. B. E. ; Poet-President, 
The Illuminating Engineering Society; Vice-President, The Illuminating 

Engineering Society (London) 

Thoroughly Revised, Enlarged and Reset 




Copyright, 1912, 



Printed by 

The Maple Press 

York, Pa, 

166353 C 

MJG 2 ^ 





Since the first edition of this book was published profound and 
revolutionary changes have taken place in the available materials 
of artificial illumination. Among electrical illuminants entirely 
new types of arc light have come into general use, and the carbon 
incandescent lamp is being rapidly pushed into obsolescence by 
the metallic filament lamps which now dominate electric lighting 

In the field of gas lighting, the inverted mantle burners of both 
large and small capacity, and the high pressure mantle burners, 
have pushed their way to the front and radically changed the 
conditions of economy which previously existed. Auxiliaries of 
every kind, and particularly shades and reflectors of greatly im- 
proved types, have been so multiplied as to meet almost every 
possible requirement. All these considerations have made neces- 
sary a very complete revision of the parts of this volume dealing 
with practical lighting. Moreover, the art of illuminating engi- 
neering has been enriched by a large amount of valuable experience 
within the past few years, and its principles are now founded on 
a more secure scientific basis. The general principles of the art, 
however, remain the same and its importance in practical life is 
at last being adequately appreciated. 

March, 1912. 




Pbeface vii 

I. Light and the Ete 1 

II. Principles of Color 25 

III. Reflection and Diffusion 37 

IV. Standards of Light and Photometry 52 

V. The Materials of Illumination — Illuminants of Com- 
bustion . 77 

VI. The Materials of Illumination — Incandescent Burners 99 

VII. The Electric Incandescent Lamp 116 

VIII. The Electric Arc Lamp 150 

IX. Shades and Reflectors 184 

X. Domestic Illumination 207 

XI. Lighting Large Interiors 233 

XII. Exterior Illumination 279 

XIII. Decorative and Scenic Illumination 316 

XIV. The Illumination of the Future ■ . 336 

Index 345 




While even the Esquimaux and the Patagonians use artificial 
light and all civilized peoples count it a necessity, it is seldom 
used skillfully and with proper knowledge of the principles that 
should govern its employment. Since the introduction of electric 
lights that very facility of application which gives them unique 
value has encouraged more zeal than discretion in their use. It 
is the purpose of the present volume to set forth some of the 
fundamental doctrines, optical, physiological, and aesthetic, which 
underlie the proper use of artificial illuminants, and to point out 
how they may be advantageously adapted to existing conditions. 

To begin with, there are two general purposes which character- 
ize two quite distinct branches of the art of illumination. First 
comes the broad question of supplying artificial light for carrying 
on such avocations or amusements as are extended into the hours 
of darkness. Quite apart from this is the case of scenic illumi- 
nation directed at special objects and designed to produce par- 
ticular effects or illusions. Lighting a shop or a house exemplifies 
the one, lighting a picture gallery or the stage of a theater the 
other. Each has a distinct purpose, and requires special means 
for its accomplishment. Confusing the purposes or mixing the 
methods often leads to serious mistakes. Sometimes both gen- 
eral and scenic illumination have to be used coincidently, but the 
distinction between them should be fully realized even when it 
cannot fully be preserved. 

General illumination, whether intended to serve the ends of 

work or play, must fulfill the following conditions: it must be 

amply adequate in amount, suitable in kind, and must be so 

applied as not to react injuriously upon the eye. 

It must be remembered that the human eye is not merely 



a rather indifferent optical instrument, but a physical organ 
which has through unfathomable ages accumulated the characters 
wrought upon it by evolution, until it bears the impress and 
incurs the limitations of its environment. It works best over a 
rather limited retinal area and through a range in intensity of 
light which, although great, is yet immensely smaller than the 
range available to nocturnal creatures. It has, moreover, become 
' habituated to, and adapted to, light coming obliquely from above, 
and resents strong illumination, whether natural or artificial, 
from any other direction. It seems to be well established, for 
example, that the distress caused by the reflected glare from 
sand, or water, or snow, and the grave results which follow pro- 
longed exposure to it, are due not only to the intensity of the 
light but to the fact that it is directed upward into the eye and 
is quite insufficiently stopped by the rather transparent lower 
eyelid. Ordinary glasses are inefficient protection in this case, 
but if the lower part of the eye be thoroughly guarded little 
difficulty is found. The Alaskan Indians have 
evolved a very effective protection against snow 
blindness in the shape of leather goggles with 
the eye arranged as shown in Fig. 1. The 
eyepiece is merely a round bit of dark leather 
with a semicircular cut made for the peephole, 
_ the resulting flap being turned outward and 

8 GoRRles downward, so that the eye is fully guarded from 

the brilliant upward beams. Blackening the 
whole lower eyelid with burnt cork is stated by one distinguished 
oculist to be completely efficacious for the same reason. 

It is more than likely that the bad effects ascribed to the habit 
of reading while lying down are due largely to the unwonted 
direction of the illumination, as well as to the unusual position 
of the eye's axis. 

All these matters are of fundamental importance in planning 
any illumination to facilitate hard visual work. Their significance 
is that we are not at liberty to depart widely from the distribu- 
tion and character of natural daylight illumination. Of course, 
one realizes immediately that the eye is neither fitted nor habit- 
uated to working to advantage in anything like the full strength 
of sunlight; but its more general properties — steadiness, dominant 
wave length, downward oblique direction, wide and strong dif- 


fusion, freedom from sharp and black shadows — these must be 
followed rather closely in ordinary artificial illumination, or the 
eye, that has been taking form through a million years of sunlight 
and skylight, will resent the change. The eye is automatically 
adjustable, it is true, for wonderfully diverse conditions, but per- 
sistent and grave changes in environment are more than it can 

Now from a practical standpoint the key to artificial illumina- 
tion is found in the thoughtful contemplation of what is known 
as Fechner's law, relating to the sensitiveness of the eye to visual 
impressions. It is stated by Helmholtz substantially as follows: 
" Within very wide limits of brightness, differences in the strength 
of light are equally distinct or appear equal in sensation, if they 
form an equal fraction of the total quantity of light compared." 
That is, provided the parts of the visual picture remain of the 
same relative brightness, the distinctness of detail does not vary 
materially with great changes of absolute brightness. Now since, 
barring binocular vision, our whole perception of visible things 
depends, in the absence of color contrasts, upon differences of 
illumination, the importance of the law just stated needs little 
comment. It implies what experience proves, that within a rather 
wide range of absolute brightness of illumination our vision is 
about equally effective for all ordinary purposes. 

Fechner's law, to be sure, fails when extremely brilliant lights 
are concerned. Few persons realize, for instance, that the sun 
is twice as bright at noon as it is when still 10 to 15 degrees 
above the horizon, still less that its brilliancy is reduced more than 
a hundred-fold as its lower limb touches the horizon. Yet while 
the eye does not detect very small changes or properly evaluate 
large ones in a body so bright as the sun, the mere fact that one 
can see to work or read about equally well from sunrise to sunset 
is most convincing as to the general truth of the law. Full sun- 
light at noon is over-bright for the eye if it falls directly upon 
the work, but with half of it or less one can get along very 

All this is most important from the standpoint of artificial 
illumination, since it means that within rather wide limits of 
intensity artificial lighting remains about equally effective for most 
practical purposes. 

The actual amount of illumination necessary and desirable, the 


terms by which we measure it, and the laws that govern its 
intensity are matters of primary importance, which must now 
occupy our attention. 

To arrive at a logical determination of the amount of illumina- 
tion necessary for general or special purposes, one must turn to 
the actual properties of the eye with respect to seeing those things 
which are customarily the objects of artificial illumination. The 
fundamental fact at the basis of vision is that the eye can perceive 
within a very wide range of absolute intensity a substantially con- 
stant fractional difference of luminosity in the objects seen. This 
is the purport of Fechner's law, to which reference has already 
been made, and the fractional difference mentioned is well known 
as Fechner's fraction. Its numerical value for ordinary eyes and 
ordinary intensities of illumination is about 0.01; that is, two 
adjacent surfaces can, under ordinary circumstances, be distin- 
guished as separate, if one reflects to the eye about one per cent 
more effective light than the other. 

It is here assumed that the objects are of approximately the 
same color, so that shade perception is the chief faculty of vision 
involved. Even if the colors are somewhat different, the value 
of Fechner's fraction is not greatly altered, provided the general 
luminosity of the two surfaces remains as stated. . In fact at a little 
distance even somewhat strongly contrasted colors blend into each 
other in a way that is altogether surprising, if they approach closely 
the same general luminosity. Now, while Fechner's fraction is 
fairly constant over a wide range of intensities, it varies, as already 
stated, when one attempts to judge extremely brilliant lights; and 
also one easily realizes that as twilight deepens his power of shade 
perception is seriously impaired. 

It is this variation of Fechner's fraction which determines the 
minimum amount of artificial, or for that matter, natural light, 
necessary for clear vision so far as shade perception is concerned. 
Now, illumination sufficient to bring Fechner's fraction up to its 
normal value, that is, to get the eye into its steady state with 
respect to shade perception, is sufficient, so far as this matter 
is concerned, for good vision, and anything above such amount 
represents waste light. 

Beside Fechner's fraction, which expresses shade perception, 
another factor of equal importance enters into practical seeing. 
This second factor is visual acuity, that is, the ability to see fine 


detail, assuming strong contrast, as, for example, between type and 
the background of the page. This power of acuity is in a great 
measure independent of the power of shade perception as such, 
being determined by other physiological peculiarities of the eye. 
It is possible, for example, to find eyes of normal acuity in which 
shade perception is somewhat deficient, and vice versa. Acuity 
seems to depend on the structure of the retina and the quality 
of the eye as an optical instrument rather than upon its direct 
or secondary sensitiveness to stimulation by light. 

In order, therefore, to see things really well one must have not 
only sufficient light to bring the eye to its steady state, but suffi- 
cient also to give the eye its normal powers of acuity. The way 
in which one's power of perceiving detail decreases in dim light 
is familiar, and the variation of acuity with the intensity of 
the illumination affords an independent criterion of the necessary 
requirements in artificial lighting. Fortunately the properties of 
the eye with respect to both shade perception and acuity have 
been the subjects of many investigations, so they may be considered 
as on the whole well determined. 






> M - 







r« v 


- - 













Fig. 2. 

Fig. 2 shows graphically the relation of strength of illumination 
to shade perception and to acuity, as determined by Dr. Uhthoff, 
and Drs. Koenig and Brodhun respectively. Curves a and 6 give 
Fechner's fraction for the normal eye for intensities up to 100 
meter-candles. Curve a, the lower one, is for white light, while 
curve b is for deep-crimson light. A little inspection will show 
that for values of the illumination below 2 or 3 meter-candles 
shade perception is somewhat deficient both for white light and 
for .crimson light, while between 10 and 20 meter-candles of 


illumination both curves rapidly merge and are settling down 
to their steady value. Above 20 to 30 meter-candles they are 
practically coincident, and power of discriminating thereafter 
remains steady up to at least some thousands of meter-candles 
of intensity. 

Hence when the light reaching the eye is above 20 to 30 
meter-candles further increase is of comparatively little assistance 
to vision so far as shade perception is concerned. The other 
curves, which are for all practical purposes coincident, are the 
acuity curves for light-orange and yellowish green lights. Within 
the range of hues found in practical illuminants, color, per se, 
makes very little difference in visual acuity. The ordinates of these 
curves are in arbitrary units, since the purpose here is merely to 
analyze their shape. Their most important feature for the present 
purpose is that, while showing low acuity at a few meter-candles, 
these curves rise very slowly after reaching 20 to 30 meter-candles, 
although they continue to rise gradually beyond this point; so it 
appears that shade perception and visual acuity reach their steady 
state in the eye for all practical purposes at about the same point, 
and that this point is not far above 20 meter-candles. In other 
words, with this illumination the eye practically reaches its normal 
working condition, and beyond this point relatively little improve- 
ment can be made by providing more light. 

Something, as will be seen later, depends upon the state of 
adaptation of the eye, that is, upon the way that it has habituated 
itself temporarily to working with more or less light. For example, 
an eye which has been working with a hundred meter-candles 
illumination finds itself somewhat inconvenienced temporarily in 
going back to 25 meter-candles, while an eye habituated to work- 
ing at 10 or 15 meter-candles can do so quite comfortably and 
would be temporarily much inconvenienced by the glare of 100 
meter-candles. The chief point to be remembered in using, as we 
shall see later, this physiological basis for the estimate of suitable 
illumination is that the meter-candles specified as necessary to 
bring the eye to its normal state refer to the light which the eye 
can derive from the objects viewed, and not merely to the inten- 
sity of the light which falls upon those objects. This is quite 
another matter, since the light emitted by the objects illuminated 
and available for the purpose in hand depends upon their reflective 
power, which will hereafter be taken into consideration. Broadly, 


the illumination available for vision of an object is that incident 
upon it multiplied by its coefficient of reflection. 

The term here used to define illumination is practically self- 
descriptive. A meter-candle of illumination means merely the 
illumination a meter from a standard candle. Similarly, 2 meter- 
candles is the illumination a meter from 2 candles, and so on. 
Until very recently there has been great confusion in the meaning 
of the term "candle" used in such connection, but in this volume 
when the term "candle" is employed, the present International 
"candle," the origin of which will be explained in Chapter IV, is 
the one thing meant. "Candle" when used in this book is used 
in this sense only. For scientific purposes the metric system is 
standard the world over, and no other system of units than the 
metric has common currency for technical purposes; hence, so 
far as the scientific investigation of illumination goes, the meter- 
candle just referred to will be employed in this volume. 

Both in England and in this country the common unit of il- 
lumination referred to in the technical press is the foot-candle, 
rather than the meter-candle, — a unit of illumination, the deri- 
vation of which is obvious in view of what has been stated regard- 
ing the meter-candle. The writer will not hesitate to use this 
common term, the foot-candle, whenever it seems desirable in 
connection with practical computation of illumination in which 
the distances rather generally are most conveniently obtained in 
feet. The illumination a foot from a candle is written both "foot- 
candle" and "candle-foot," the latter term being common in 
English books. The terms are absolutely interchangeable, and the 
use of both of them can create no confusion, although the writer 
personally prefers and uses the former mainly on account of its 
more euphonious and descriptive plural. The relation between 
these practical units of illumination is very simple: 1 foot-candle 
equals 10.76 meter-candles, so that no confusion need result from 
the double use of terms. As will be scfen later, the meter-candle is 
the systematic unit of illumination to which properly belongs the 
name lux. 

For any light the illumination at one meter distance is obviously 
a number of meter-candles numerically equal to the candle power 
of the light. 

At distances other than one meter the illuminating power is de- 
termined by the well-defined, but often misapplied, "law of inverse 



squares." This law states that the intensity of light from a 
given source varies inversely as the square of the distance from 
that source. Thus, if we have a radiant point (P, Fig. 3), it will 
shine with a certain intensity on a surface abed at a distance eP. 
If we go to double the distance (EP), the same light which fell on 
abed now falls on the area ABCD } of twice the linear dimensions 
and four times the area, and consequently the intensity is reduced 
to one-fourth of the original amount. Thus if P be one candle 
and eP one meter, then the illumination at e will be one meter- 
candle, and at E one-fourth meter-candle. 

This law of inverse squares is broadly true of every case of the 
free distribution of energy from a point within a homogeneous 
medium, for reasons obvious from the inspection of Fig. 3. > It does 

Fig. 3. — Illustrating Law of Inverse Squares. 

not hold save within certain limits in considering a radiant surface 
as a whole, nor for any case in which the medium is not homo- 
geneous within the radii considered. 

By reason of these limitations, in problems of practical illu- 
mination the law of inverse squares can be considered only as a 
useful guide; for it is far from infallible, and may lead to grossly 
inaccurate results. It is exact only in the rare case of radiation 
from a minute point into space in which there is no refraction 
or reflection. A room with dead-black walls, lighted by a single 
candle, would furnish an instance in which the illumination could 
be computed by the law of inverse squares without an error of 
more than say 2 or 3 per cent, while a whitc-and-gold room lighted 
by a well-shaded arc light would illustrate an opposite condition in 
which the law of inverse squares alone would give a result grossly 
in error. 


Fig. 4 shows how completely deceptive the law of inverse squares 
may become in cases complicated by refraction or reflection. Here 
one deals with an arc light of perhaps 5000 actual candle power as 
the source of radiation, but a very large proportion of the total 
luminous energy is concentrated by the reflector or lens system into 
a nearly parallel beam which maintains an extremely high lumi- 
nous intensity at great distances from the apparatus. If the beam 
were actually of parallel rays its resultant illumination would be 
uniform at all distances, save as diminished by the absorption of 

Fig. 4. — Beam from Searchlight. 

the atmosphere, probably not over 10 per cent in a mile in ordi- 
narily clear weather, since the absorption of the entire thickness of 
the atmosphere for the sun's light is only about 16 per cent. 

The searchlight furnishes really a special case of scenic illumina- 
tion, which frequently depends upon the use of concentrated beams 
in one form or another, so that one must realize that a very con- 
siderable branch of the art of illumination imposes conditions not 
reconcilable with the ordinary application of the law of inverse 

It is worth while thus to examine the law in question because 
it is a specially flagrant example of a principle, absolutely and 


mathematically correct within certain rigid limitations, but par- 
tially or wholly inapplicable in many important cases. 

Aside from the lux, which is little used in this country as com- 
pared with the foot-candle, there is only one generally accepted 
special unit employed in illuminating engineering. This is based 
upon the idea of luminous flux, that is, luminous energy proceed- 
ing from a point into the surrounding free space. Evidently such 
luminous flux determines the whole quantity of light energy which 
streams from a single source, diminishing in flux density per unit 
area as it proceeds outward. The total luminous energy evidently 
remains the same, whatever the total illuminated area around the 
source may be. Precisely this idea of flux runs through all cases 
of energy outflowing from a central source. 

The unit of luminous flux is taken as that proceeding from a 
source of unit intensity throughout one unit solid angle, and is 
called the lumen. There are 4x or 12.56 such solid angles in the 
sphere. Now if one takes the international candle as the unit of 
intensity, then 1 lumen is the flux of luminous energy proceeding 
from a source of 1 mean spherical international candle through a 
unit solid angle; and in terms of flux, therefore, 1 mean spherical 
international candle is a source of 4 v lumens. As the luminous 
flux proceeding outward is not increased or decreased in total 
amount at whatever distance it is measured, any surface sub- 
tending one unit solid angle from the source mentioned receives a 
total flux of 1 lumen. This total flux, divided by the area of the 
surface in square feet, gives the illumination in foot-candles at that 
surface; or, if one chooses the lux as the unit of illumination, the 
total flux over the area must be divided by the area in square 
meters to give the corresponding- illumination. The foot-candle, 
therefore, denotes an illumination of 1 lumen per square foot and 
the lux an illumination of 1 lumen per square meter. This method 
of reckoning provides a very convenient way of getting the illu- 
mination, provided one knows or can compute the efficiency of 
utilization of the source reckoned upon the working plane. A 
table of such efficiency with various kinds of installations will 
be given later which will prove useful in computing the necessary 
intensity of the source to produce a given illumination in lumens 
per square foot or per square meter. Perhaps the most important 
use of the idea of luminous flux is in reckoning the illumination 
proceeding from secondary sources like bright illuminated surfaces. 


The lumens incident upon these can be at once computed from 
the solid angle subtended by them with respect to the primary 
source. This determined, the secondary source becomes simply a 
source of a known number of lumens at a given distance from the 
point at which the secondary illumination is to be reckoned.* 

Several systems of units connected with illumination have been 
from time to time proposed, but have not, save for the lux and 
the lumen, which are common to all of them, met with sufficient 
general acceptance to render discussion of them here profitable. 
Most actual computations of illumination are made on the basis 
of the intensity of the sources and their relation to the surfaces 
to be illuminated, or by the flux-of-light method referred to the 
efficiency of utilization on the working plane. 

Having considered the unit strength of light and the unit strength 
of illumination and of luminous flux, the other fundamental of 
artificial lighting is the intensity of the luminous source — gene- 
rally known as intrinsic brightness. Optically this has no very 
great or direct importance, but physiologically it is of the most 
serious significance, and perhaps deserves more thoughtful atten-. 
tion than any other factor in practical illumination. It is of the 
more consequence, as it is the one thing which generally receives 
scant consideration, and is left to chance or convenience. 

By intrinsic brightness is meant the strength of light per unit 

area of light-giving surface. If we adopt the standard candle as 

the unit of light, and adhere to English measures, the logical unit 

of intrinsic brightness is one candle power per square inch. One 

then may conveniently express the brightness of any luminous 

surface in candle power per square inch, and thus obtain a definite 

basis of comparison, as in the accompanying table. 

* An interesting modification of the flux-of-light method of reckoning illu- 
mination is the absorption method of Dr. McAllister (Electrical World , Nov. 
21, 1908). This is based upon the fact that whatever the intensity of illu- 
mination in, for instance, a room, for that intensity the light sources must 
produce the sum of all the luminous flux absorbed at the surfaces. Now 
the light-absorption coefficient is the familiar quantity (1 — k) } and for a re- 
quired flux density in foot-candles the necessary lumens equal this flux den- 
sity multiplied by the area and by (1 — k). Hence whenever these quantities 
are known for the various surfaces considered the total lumens, and hence 
the required candle power can be at once ascertained. This very ingenious 
method, which is, so to speak, the converse of the ordinary flux-of-light 
computation, is occasionally very useful, and its details may be found in the 
highly original paper to which reference has been made. 




Source. Brilliancy. Notes. 

Sun in zenith. •• — •■• • J^'^n I Rou « h equivalent values, taking ac- 
Sun at 30 deg. elevation. . 600,000 > co * £ absorption. 

Sun on horizon 2,000 i ^ 

Arc light 10,000 to 100,000 

Flame arc 5,000 Clear globe. 

Calcium light 5,000 

Magnetite arc 4,000 

Nernst "glower" 3,000 Unshaded. 

Tungsten lamp 1,000-1,100 

G. KM. lamp 750 

Quartz Mercury arc 600-1,000 

Tantalum lamp 580 

Caibon incandescent lamp 300-500 Depending on efficiency. 

Melting platinum 129 

Inclosed arc 75-150 Opalescent inner globe. 

Acetylene flame 40-60 

Welsbach light 20 to 40 j "j^y™ ™ for hl 8 h "P re8sur e 8^ 

Cooper Hewitt tube 10 to 12 

Kerosene light 4 to 9 Very variable. 

Candle 3 to 4 

Gas flame 3 to 8 Very variable. 

Incandescent (frosted) . . 2 to 8 
Opal-shaded lamps, etc. . 0.5 to 2 
Moore tube 0.5 to 1 

The striking thing about this table is the enormous discrepancy 
between electric and other lamps of incandescence and flames of 
the ordinary character. The very great intrinsic brilliancy of the 
ordinary unshaded incandescent lamps is particularly noteworthy 
and, from the oculist's standpoint, menacing. 

Although a measure of intrinsic brightness is obtained by divid- 
ing the candle power of any light by the area of the luminous 
surface, this latter quantity is very difficult to determine accu- 
rately, since with the exception of the electric incandescent filament 
no source of light is anywhere nearly of uniform brilliancy over 
its entire surface. For the sake of comparison we can, however, 
draw up the above approximate table by assuming equal bright- 
ness over the generally effective lighting area of any radiant. It 
should be distinctly understood that the values tabulated are only 
average values of quantities, some of which are incapable of 
exact determination and others of which vary over a wide range 
according to conditions. 

Everyone is familiar with the distress caused the eye by sud- 
den alternations of light and darkness, as in stepping from a dark 
room into full sunlight, or even in lighting the gas after the eye 


has become habituated to the darkness. The eye is provided with 
a very wonderful automatic "iris diaphragm' ' for its adjustment 
to various degrees of illumination, but- it is by no means instan- 
taneous, although very prompt, in its action. Moreover, the eye 
after resting in darkness is in an extremely sensitive and receptive 
state, and a relatively weak light will then produce very noticeable 
after-images. These after-images, such as are seen in vivid colors 
after looking at the sun, are due to retinal fatigue. 

If the image of a brilliant light is formed upon the retina, it 
produces certain very considerable chemical changes, akin to those 
produced by light upon sensitized paper. In so doing it tempo- 
rarily exhausts or weakens the power of the retina to respond at 
that point to further visual impressions, and when the eye is turned 
away the image appears, momentarily persistent, and then reversed, 
dark for a white image, and of approximately complementary hue 
for a colored one. This after-image changes color and fades away 
more or less slowly, according to the intensity of the original 
impression, as the retina recovers its normal sensitiveness. 

A strong after-image means a serious local strain upon the eye, 
and shifting the eye about when brilliant light can fall upon it 
implies just the same kind of strain that one gets in going out of 
a dark room into bright sunshine. The results may be very seri- 
ous. In one case recently reported a strong side light from an 
unshaded incandescent lamp set up an inflammation that finally 
resulted in the loss of an eye. The light was two or three feet 
from the victim, whose work was such that the image of the 
filament steadily fell on about the same point on the retina, at 
which point the resulting inflammation had its focus. A few 
weeks' exposure to these severe conditions did the mischief. This 
is an extreme case, but similar conditions may very quickly cause 
trouble. A few years ago the writer was at lunch facing a window 
through which was reflected a brilliant beam from a white-painted 
sign in full sunlight just across the street. No especial notice was 
taken of this, until on glancing away a strong after-image of the 
sign appeared, and although the time of exposure was only ten 
or fifteen minutes, the net result was inability to use the eyes more 
than a few minutes at a time for a fortnight afterwards. 

To a certain extent the eye can protect itself from the bril- 
liant sources of light by the automatic action of the iris. This 
protection, however, is not rapid enough or complete enough to 



guard the eye properly against the brilliant sources now in com- 
mon use. 

It is very difficult to get an exact idea of the reaction of the 
pupil to light on account of the large number of factors which enter 
the question and the constant slight variations to which the pupil- 
lary diameter is subject. Its diameter varies from scarcely more 
than 1 mm. under extreme conditions of contraction, to 7 or 8 mm. 
in darkness, so that to use the familiar expression applied to lens 

/ / 
stops, it works from somewhere about -~ to ^4 > 

10 z.o 

or even | , when 




51 30 













Fig. 5. — Variation of the Pupil in Different Illuminations. 
Plotted from early experiments by Lambert. 

in darkness the expanding iris retreats clear out to the rim of the 
cornea. Ordinarily the pupillary diameter is in moderate light 

3 or 4 mm., and the eye therefore is working at about j • 

A rough idea of the variation of the pupil in different illumi- 
nations is given by the curve of Fig. 5, plotted from the early 
experiments of Lambert. The ordinates give the area of the 
pupil, the abscissae the illumination, in meter-candles. It is 
interesting to note that most of the variation takes place under 


10 meter-candles, beyond which the curve rapidly becomes asymp- 
totic. The eye cannot, therefore, well protect itself against ex- 
tremely bright sources, and seems in this, as in other particulars, 
to have been specialized in the course of its development for 
moderate degrees of illumination; nor is the protection instan- 
taneously established. It takes about half a second for contrac- 
tion or expansion to set in after a sudden change in illumination. 
The contraction, once begun, takes, however, less than half this 
time, and expansion somewhat longer. The eye, therefore, cannot 
effectively guard itself against sudden variations, and the result 
is often extremely painful. 

An important question is the effect upon the pupil of such dis- 
tribution of light as is commonly found in artificial illumination. 
Does the pupil adjust itself to the average intensity or to the 
intensity of the brightest point within the field of vision? This 
matter has been pretty thoroughly investigated, with the result 
of showing that upon the whole the pupil adjusts itself rather 
to bright lights in the central part of the field than to average 
illumination. It does not, however, react as fully to bright lights 
in the peripheral field, and thus defends itself rather inadequately 
against intense light coming from unwonted directions. 

The presence of a bright light in full view, therefore, causes the 
pupil to contract, and seriously reduces the visibility of objects in 
the adjacent field. In ordinary seeing, where there are no brilliant 
sources visible, the iris opens up when the lighting is low and gives 
considerably increased powers of discrimination. Were it not for 
this, it would be exceedingly difficult to get about at night even 
by moonlight. In this latitude; moonlight even near full moon is 
hardly more than 0.2 meter-candles, which by reference to Fig. 2 
would give Fechner's fraction at nearly 0.5, save for the aid 
received from the expanding pupil. With the pupillary area, how- 
ever, increased perhaps six times, one can see to get about com- 
fortably enough and can even read very coarse print. It should 
be noted here that the curves of Fig. 2 were attained by vision 
through a stop, so that the effective pupillary diameter was sensibly 

The same conditions have an important bearing on vision in 
presence of a brilliant light in the field. For example, suppose 
that in a general illumination of 1 meter-candle the eye can make 


out objects having a contrast -j equal to 0.15. Then let a light 


come into the field of vision so as to increase the illumination on 
the eye to 20 meter-candles without materially illuminating the 
objects in the vicinity. The pupil will close to about one-third of 
its former area under these circumstances, raising Fechner's fraction 
to 0.3 or thereabouts, and consequently objects having the contrast 
just mentioned would disappear. 

Hence, as is well known, one cannot see well across a bright light, 
and even objects illuminated by it will lose in visibility unless 
the change in the illumination received by them is greater than the 
concomitant adverse change produced by the contraction of the 
pupil. In short, a bright light falling on the eye quite generally 
interferes with vision by decreasing the pupillary aperture, more 
than it helps it by added illumination upon neighboring objects. 

A very simple experiment, showing this effect of a strong source 
of light on the apparent illumination, may be tried as follows: 
Light a brilliant lamp, unshaded, in a good-sized room, preferably 
one with darkish paper. Then put on the light an opal or similar 
shade. It will be found that the change has considerably im- 
proved the apparent illumination of the room, although it has 
really cut off a good part of the total light. Moreover, at points 
where there remains a fair amount of illumination, the shade has 
improved the reading conditions very materially. If one is reading 
where the unshaded light is at or within the edge of the field of 
vision, the improvement produced by the shade is very conspicu- 
ous. Lowering the intrinsic brilliancy of the light has decreased 
the strain upon the eye and given it a better working aperture. 

As a corollary to these suggestions on the effect of bright lights 
on our visual apparatus should be mentioned the fact that sudden 
variations in the intensity of illumination seriously strain the eye 
both by fatigue of the retina, due to sudden changes from weak to 
strong light, and by keeping the eye constantly trying to adjust 
itself to changes in light too rapid for it properly to follow. 

A flickering gaslight, for example, or an incandescent lamp run 
at very low frequency, strains the eye seriously and is likely to 
cause temporary, even if not permanent, injury. 

The persistence of visual impressions whereby the retinal image 
remains steady for an instant after the object ceases to affect the 
eye furnishes a certain amount of protection in case of very rapid 
changes of brilliancy. It acts like inertia in the visual system. 

In the case of arc and incandescent lamps, the thermal inertia of 


the filament or carbon rod also tends physically to minimize the 
changes, but with a low-frequency alternating current they may 
still be serious. 

The exact frequency at which an incandescent lamp on an alter- 
nating circuit begins to distress the eye by the flickering effect de- 
pends somewhat on the individual eye and somewhat on the mass 
of the filament. In general, a 16-c.p. lamp of the usual voltages, 
say 100 to 120 volts, begins to show flickering at or sometimes a 
little above 30 cycles per second; one foreign authority noting it 
even up to 40 cycles. At 25 cycles the flickering is troublesome to 
most eyes, and at 20 cycles or below it is generally quite intoler- 
able. In looking directly at the lamp the filament is so dazzling 
that the fluctuations are not always in evidence at their full value, 
and a low-frequency lamp is quite likely to be the source of trouble 
to the eye even when at first glance it appears to be quite steady. 
The metallic filament lamps from their small thermal inertia are 
more sensitive to these effects than carbon-filament lamps. 

Lamps having relatively thick filaments can be worked at lower 
frequencies than those of the common sort, so that 50-volt lamps, 
particularly of large candle power, may be worked at 30 cycles or 
thereabouts rather well, and out of doors even down to 25 cycles. 
That is, at a pinch one can do satisfactory work when current is 
available at 25 cycles or so, by using low-voltage lamps of 32, 
50, or 100 candle power, which, by the way, are capable of giving 
admirable results in illumination if properly disposed. Of course, 
such practice is bad in point of efficient distribution of current, 
but on occasion it may be useful. 

As to arc lamps, conditions are not so favorable. The fluc- 
tuations of an alternating arc lamp are easily detected, even at 
60 cycles, by moving a pencil or the finger quickly when strongly 
illuminated. The effect is a series of images along the path of 
motion, corresponding to the successive maxima of light in the 
arc. At 40 to 45 cycles the flickering becomes evident even when 
viewing stationary objects, the exact point where trouble begins 
depending upon the adjustment of the lamp, the hardness of the 
carbons, and various minor factors. Inclosing the arc mitigates 
the difficulty somewhat, but does not remove it. 

In working near the critical frequency the best results are 
attained by using an inclosed arc lamp taking all the current 
the inner globe will stand, with as short an arc as will work 


steadily. Flaming arcs perform rather better on account of the 
large mass of light-giving vapor. 

When polyphase currents are available, as is usually the case 
where rather low frequencies are involved, some relief may be 
obtained by arranging the arcs in groups consisting of one from 
each phase. At a little distance from such a group the several 
illuminations blend so as to partially suppress the fluctuations 
of the individual arcs. This device makes it possible to obtain 
fairly satisfactory lighting between 35 and 40 cycles. At these 
frequencies, however, the arcs should not be used except when 
a very powerful light is necessary, or when the slightly yellowish 
tinge of incandescents would interfere with the proper judgment 
of colors. Powerful incandescents are generally better, and are, 
now that large tungsten lamps are available, quite as efficient, 
particularly when one takes into account proper distribution of 
the light. In using incandescents in large masses, particularly 
on polyphase circuits, the flickering of the individual lights is 
lost in the general glow, so that even at 25 cycles the light may 
be steady enough for general purposes, as was the case with the 
decorative lighting at the Pan-American Exposition. The fluc- 
tuations due to low frequency are usually very distressing to the 
eye, and should be sedulously avoided. Fortunately, save in rare 
instances, the frequency can be and should be kept well above 
the danger point. 

The same considerations which forbid the use of very intense 
lights, unshaded; flickering lights; and electric lights at too low 
frequency, render violent contrasts of brilliant illumination and 
deep shadows highly objectionable. It should be remembered that 
in daylight the general diffusion of illumination is so thorough 
that such contrasts are very much softened, even in full sunlight, 
and much of the time the direct light is modified by clouds. In 
situations where the sun shines strongly down through interstices 
in thick foliage, the effect is decidedly unpleasant if one wishes 
to use the eyes steadily; and if, in addition, the wind stirs the 
leaves and causes flickering, the strain upon the eyes is most 

In artificial lighting one should carefully avoid the conditions 
that are objectionable in nature, which can easily be done by 
a little foresight. If for any purpose very strong illumination 
becomes necessary at a certain point, the method of furnishing it 


which is most satisfactory from a hygienic standpoint is to super- 
impose it upon a moderate illumination well distributed. If a 
brilliant light is needed upon one's work, start with a fairly well- 
lighted room and add the necessary local illumination, instead of 
concentrating all the light on one spot. This procedure avoids 
dense shadows and dark corners, and enables the eye to work 
efficiently in a much stronger illumination than would otherwise 
be practicable. 

It should not be understood that the complete abolition of 
shadows is desirable. On the contrary, since much of our percep- 
tion of form and position depends upon the existence of shadows, 
the entire absence of them is troublesome and unpleasant. This is 
probably due to two causes. First, the absence of shadows gives 
an appearance of flatness out of which the eye vainly struggles to 
select the wonted degrees of relief. In a shadowless space we 
have to depend upon accommodationand binocular vision to locate 
points In three dimensions, and the strain upon the attention is 
severe and quickly felt. 

Second, the existence of a shadowless space presupposes a nearly 
equal illumination from all directions. If it be strong enough 
from any particular direction to be convenient for work requiring 
close attention of mind and eye, then, if there be no shadows, 
equally strong light will enter the eye from directions altogether 
unwonted. This state of things we have already found to be 
objectionable in the highest degree. 

The best illustration of this unpleasant condition may be found 
in nature during a thin fog which veils the sun while diffusing 
light with very great brilliancy. Try to read at such a time out of 
doors, and, although there is no direct light on the page to dazzle 
you, and there is in reading no trouble from the sense of flatness, 
yet there is a distinctly painful glare which the eyes cannot long 
endure without serious strain. 

In artificial lighting the same complete diffusion is competent 
to cause the same results, so that while contrasts of dense shadows 
and brilliant light must be avoided, it is generally equally impor- 
tant to give the illumination, even if deliberately indirect, a certain 
general direction to relieve the appearance of flatness and to save 
the eye from cross lights. 

With respect to the best direction of illumination, only very 
general suggestions can be given. Brilliant light, direct or re- 


fleeted, should be kept out of the eye and upon the objects to be 
illuminated. In each individual case the nature and requirements 
of the work must determine the direction of lighting. 

The old rule given for reading and writing, that the light should 
come obliquely over the left shoulder, well illustrates ordinary 
requirements. By receiving the light from the point indicated 
direct light is kept out of the eyes, and any light regularly re- 
flected is generally out of the way. The eye catches then only 
diffused light from the paper before it, and if the light comes from 
the left (for a right-handed person) the shadow of hand and arm 
does not interfere with vision in writing. If work requiring both 
hands is under way, the chances are that the best illumination will 
be obtained by directing it downwards and slightly from the front, 
in which case care must be exercised to avoid strong direct reflec- 
tion into the eyes. The best simple rule is, avoid glare direct or 
reflected, avoid strong shadows, and get ample diffused light from 
the object illuminated. 

This brings us at once to the very important but ill-defined 
question of the strength of illumination required for various kinds 
of work. 

Fortunately, the eye works well over a wide range of brightness, 
but there is a certain minimum illumination which should be ex- 
ceeded if one is to work easily and without undue strain. The 
matter is much complicated by questions of texture and color, 
which will be taken up presently, so that only general average 
results can be considered. For reading and writing, experience 
joins the physiological data already given in showing that an in- 
tensity of at least 10 meter-candles is the minimum amount for 
ordinary type and ink, such as is here used, for instance. With 
large, clear type, 

like that used for this particular line, 

5 or 6 enable one to read rather easily; while with ordinary type 
set solid or in type of the smaller sizes, 

such type aa is employed in this line as a horrible example, 

30 or 40 meter-candles is by no means an unnecessary amount of 
lighting. Dense black ink and clear white paper not highly calen- 
dered, such as some of the early printers knew well how to use, 
make vastly easier reading than the grayish-white stuff and cheap, 
muddy-looking ink to be found in the average newspaper. 


Illumination of less than 10 usually renders reading somewhat 
difficult and slow, the more difficult and slower as the illumination 
is further reduced. At 2 or 3 meter-candles reading is by no means 
easy, and there is a strong tendency to bring the book near the 
eye, thereby straining one's power of accommodation, and to con- 
centrate the attention upon single words, a tendency which in- 
creases as the light is still further lessened. 

In fact, when the illumination falls to the vicinity of 1 meter- 
candle it is of very little use for the purpose of reading or 

One may get a fair idea of the strength of illumination required 
for various purposes by a consideration of that actually furnished 
by Dature. To get at the facts in the case, we must make a little 
digression in the direction of photometry, a subject which will be 
more fully discussed later. 

To get an approximate measure of the illumination furnished 
by daylight, one can conveniently use what is known as a daylight 
photometer. This instrument furnishes a means for balancing the 
illumination due to any source against that due to a standard 
candle at a known distance. Like most common forms of photom- 
eter, it consists of a screen illuminated on its two sides by the 
two sources of light respectively. Equality of illumination is de- 
termined by the disappearance of a grease spot upon the screen. 
A spot of grease on white paper produces, as is well known, a 
highly transparent spot, which looks bright if illuminated from 
behind, and dark when illuminated from the front. 

Thus, if one sets up such a screen C between, and equi-distant 
from, a candle A and an incandescent lamp B, and then looks at 
the screen obliquely from the same side as B, the appearance is 
that shown in Fig. 6. Moving around to the other side of the 
screen, one gets the effect shown in Fig. 7. By moving the candle 
A nearer or the incandescent B farther off, a point will be found 
where the spot becomes nearly invisible on account of the equal 
illumination on the two sides. This " Bunsen photometer screen " 
requires very careful working to get highly accurate results, but 
gives closely approximate figures readily. The daylight photom- 
eter, Fig. 8, is the simplest sort of adaptation of this principle. 
It consists of a box, say 5 or 6 feet long and 15 inches square. 
In one end is a hole B filled with the photometer screen just 
described, and a slot to receive a graduated scale A carrying a 



socket for a standard candle. The interior of the box is painted 
dead black, so as to avoid increasing the illumination at B by light 
reflected within the box. 

Fig. 6. — Principle of the Photometer. 

Setting up the box with the end B pointing in the direction of 
the illumination to be estimated, the candle is slid back and forth 
until the grease spot disappears, when the distance from the candle 
to B gives the required illumination, by applying the law of 



Fig. 7. — Principle of the Photometer. 

inverse squares, which holds sufficiently well for approximate 
purposes if the box is well blackened. 

Of course the results of such measurements vary enormously 
with different conditions of daylight. A few measurements made 
in a large, low room with windows on two sides, culled from the 


writer's notebook, give the following results, the day being bright, 
but not sunny, and the time early in the afternoon: 

Facing south window 64 meter-candles. 

Facing east window 24 " " 

Facing north wall 7.5 " " 

And again, 10 feet from south window, on a 

misty April day, 6 P.u 5.3 " " 

On a clear day the diffused illumination near a window, while 
the sun is still high, will generally range from 50 to 60 meter- 
candles, while in cases where there are exceptionally favorable 
conditions for brilliant illumination it may rise to twice or even 
four times the amount just stated. The intrinsic brilliancy of an 
aperture fully exposed to the upper sky is, for a yearly average, 
according to the measurements of Dr. Basquin, about 0.4 candle 
power per square centimeter, which enables the illumination to be 
roughly estimated in simple cases. 

Fig. 8. — Daylight Photometer. 

Now, these figures for the lighting effects of diffused daylight 
give a good clew, 'if nothing more, to the intensity of illumination 
required for various purposes. In point of fact, reading and 
writing require less light than almost any other processes which 
demand close ocular attention. Everything is black and white, 
there is no delicate shading of colors, nor any degrees of relief 
to be perceived in virtue of differences of light and shade. More- 
over, the characters are sharply defined and not far from the eye. 
It is therefore safe to say that for even the easiest work requiring 
steady use of the eyes at least 10 meter-candles are demanded. 
In general, this minimum should be at least doubled for really 
effective lighting, while for much fine detail and for work on 
colored materials not less than 50 meter-candles should be pro- 


vided. Even this amount may advantageously be doubled for the 
finest mechanical work, such as engraving, watch repairing, and 
similar delicate operations. In fact, for some such cases the more 
light the better, provided the source of light and direct undiffused 
reflections therefrom are kept out of the eyes. 

These estimates have taken no account of the effect of color, 
which sometimes is a most important factor, alike in determin- 
ing the amount of illumination necessary and in prescribing the 
character and arrangement of the sources of light to be employed. 



The relation of color to practical illumination is somewhat 
intricate, for it involves considerations physical, physiological, 
and aesthetic; but it is well worth studying, for while in some 
departments of illumination, such as street lighting, it is of little 
consequence, in lighting interiors it plays a very important part. 
In lighting a shop where colored fabrics are displayed, for exam- 
ple, it is necessary to reproduce as nearly as may be the color 
values of diffused daylight, even at considerable trouble. Such 
illumination, however, may be highly undesirable in lighting a 
ballroom, where the softer tones of a light richer in yellow and 
orange are generally far preferable. 

In certain sorts of scenic illumination strongly colored lights 
must be employed, but always with due understanding of their 
effect on neighboring colored objects. Sometimes, too, the nat- 
ural color of a light needs to be slightly modified by the presence 
of tinted shades, serving to modify both the intrinsic brilliancy 
and the color. 

The fundamental law with respect to color is as follows: Every 
opaque object assumes a hue due to the sum of the colors which it 
reflects. A red book, for instance, looks red because from white 
light it selects mainly the red for reflection, while strongly absorb- 
ing the green and blue. 

White light, as a look through a prism plainly shows, is a com- 
posite of many colors, fundamentally red, green, and blue, inci- 
dentally of an almost infinite variety of transition tints. If a 
narrow beam of sunlight passes through a prism, it is drawn 
out into a many-colored spectrum in which the three colors 
mentioned are the most prominent. Closer inspection detects 
a rather noticeable orange region passing from red to green by 
way of a narrow space of pure yellow, which is never very con- 
spicuous. The green likewise shades into pure blue through a 
belt of greenish blue, and the blue in turn shades off into a 
deep violet. If the slit which admits the sunlight is made very 



narrow, certain black lines appear crossing the spectrum, — the 
Fraunhofer lines due to the selective absorption of various sub- 
stances in the solar atmosphere. These lines are for the purpose 
in hand merely convenient landmarks to which various colors 
may be referred. They were designated by Fraunhofer by the 
letters of the alphabet, beginning at the red end of the spectrum. 
Fig. 9 shows in diagram the solar spectrum with these lines 
and the general distribution of the colors. The A line, really 
a broad dark band of many lines, is barely visible save in the 
most intense light, and the eye can detect little or nothing 
beyond it. At the other end of the spectrum the H lines are in 

Fig. 9. — Solar and Reflected Spectra. 

a violet merging into lavender, are not easy to see, and there is 
but a narrow region visible beyond them, — pale lavender, as 
generally seen. The spectrum in Fig. 8 is roughly mapped out 
to show the extent of the various colors as distributed in the 
ordinary prismatic spectrum. 

At A, Fig. 9, is shown the spectrum of the light reflected from 
a bright-red book, i.e., the color spectrum which defines that 
particular red. It extends from a deep red into clear orange, 
while the absorption in the yellow and yellowish green is by no 
means complete. 

At B is the color spectrum from a green book. Here there 
is considerable orange and yellow, a little red and much bright 
green, together with rather weak absorption in the bluish green. 


C shows a similar diagram from a book apparently of a clear, 
full blue. The spectrum shows pretty complete absorption in tbe 
red and extending well into the orange. The orange-yellow and 
yellowish green remain, however, as does all the deep blue, while 
there is a perceptible absorption of the green and bluish green. 

Now, these reflected spectra are thoroughly typical of those 
obtained from any dyed or painted surfaces. The colors ob- 
tained from pigments are never the simple hues they appear 
to be, but mixtures more or less complex sometimes of colors 
from very different regions of the spectrum. Most of tbe com- 
moner pigments produce absorption over rather wide regions of 
the spectrum, but some of the delicate tints found in dyed fabrics 
show several bands of absorption in widely separated portions 
of the spectrum. These arc the colors most seriously affected 
by variations in the color of the illuminant when viewed by 

Fig. 10. — Spectrum Reflected from Blue Silk. 

artificial light. Fig. 10 is a case in point, a color spectrum taken 
from a fabric which in daylight was a delicate cornflower blue. 
The absorption begins in the crimson, leaving much of the red 
intact, is partial in the orange and yellow, stronger in the green, 
and quite complete in tbe bluish-green region. The blue well up 
to the violet is freely reflected, and then the violet end of the 
spectrum is considerably absorbed. Most of the reflected light 
is blue, but if the illumination is conspicuously lacking in blue 
rays, as is the case with candlelight or common gaslight, the blue 
light reflected is necessarily weak, while the red component comes 
out at its full strength, and the visible color of the fabric is dis- 
tinctly reddish. 

A similar condition is met in certain blues which in daylight 
reflect a large proportion of blue and bluish violet, but in which 
some green rays are left, just as was the clear red in Fig. 10. By 
gaslight the blue becomes relatively very much weakened, and the 
apparent color is unmistakably green. Such changes in hue are in 



greater or less degree very common, and furnish some very curi- 
ous effects. Sometimes a color clear by daylight appears dull and 
muddy by artificial light, and in general the quality of the illumi- 
nation requires careful attention whenever one deals with delicate 

The absorption found in the pigments used in painting is seldom 
so erratic as that shown in Fig. 10, but pictures often show very 
imperfectly under ordinary artificial illumination. 

It is no easy matter to get a clear idea of the color properties 
of various illuminants. Of course, one can form spectra from 
each of the lights to be compared, and compare the relative 
strengths of the red, green, blue, and other rays in each; but this 
gives but an imperfect idea of the relative color effects produced, 
for the results themselves are rather discordant, and the relative 
brightness thus measured does not correspond accurately with the 
visual effect. Lights have also been extensively compared by 
color-mixing devices using colored screens to segregate red, green, 
and blue portions of the spectrum which are then varied to match 
the color under investigation. The results are valuable inter se, 
but lack the definiteness secured by using the spectral colors. 
Probably a better plan from the standpoint of illumination is to 
match the visible color of a given illuminant accurately by mix- 
tures of the three primary spectral colors, red, blue-violet, and 
green, and to determine the exact proportions of each constituent 
required to give a match. Even this evidently does not tell the 
whole story, but it gives an excellent idea of the color differences 
found in various lights. Such work has been very beautifully 
carried out by Abney, from whose results the following table is 



Sky light. 

Arc light. 









Incandescent lamps are not here included, but give enormously 
different results according to the degree of incandescence to which 
they are carried. If burned below candle power, they give a light 
not differing widely from gaslight; while if pushed far above candle 
power, the light is far richer in violet rays, and becomes approxi- 


mately white. Unfortunately, however, the lamp does not reach 
this point save at a temperature that very quickly ends its life. 

The effects of the selective absorption which so deceives the 
eye when colored objects are viewed in colored lights are shown 
in a variety of ways according to the colors involved, but the net 
result of them all is to show the necessity of looking out for the 
color of artificial lights. Of course, a really strong color may 
produce very fantastic results. For example, in the rays of an 
ordinary green lantern, such as is used for railway signals, greens 
generally appear of nearly their natural hues; but greens, yellows, 
browns, and grays all match pretty well, although they may appear 
darker or lighter in shade. Pink looks gray, darkening in shade 
as it is redder, and red is nearly black, for the green light which 
falls upon it is almost totally absorbed. 

Practical illuminants do not often present so violent deceptions, 
and yet gas or candle light is certain to change the apparent hue 
of any delicate colors containing bluish-green, blue, or violet rays. 
An old Welsbach mantle which gives a light of a strongly greenish 
cast is pretty certain to change the color of everything not green 
upon which it falls. Incandescent electric lights affect colors in 
much the same way as brilliant gaslight, while arc lights give a 
fair approximation to daylight. It by no means follows, however, 
that all colors should be matched by arc lights in preference to 
other sources of illumination. A match so made stands daylight, 
but may be most faulty when viewed by gaslight. 

If matching colors has to be done, it is a safe rule to match them 
by the kind of light by which they are intended to be viewed. 
Moreover, different shades of the same color are differently affected 
in artificial light. As a rule, deep, full colors are far less affected 
than light tones of the same general hue. Clear yellows, reds, 
and blues not verging on green are usually little altered, but pale 
pinks, violets, and " robin's-egg ' ' blues quite generally suffer. Very 
often when a color is not positively altered it is made to appear 
gray and muddy. 

For while in a green light greens look particularly brilliant, red 
may be practically extinguished, absorbing all the rays which 
come to it, so that a deep red will be nearly black, and a very light 
red merely a dirty white, tinged with green if anything. 

Quite apart from any effect of colored illumination, colors seem 
to change in very dim light. This is a purely physiological matter, 



the eye itself differing in its sensibility to different colored lights. 
In very faint illumination no color of any kind is perceptible — 
everything appears of uncertain shades of gray. As the light fades 
from its normal intensity, as in twilight, red disappears first, 
then violet and deep blue follow, settling like the red into murky 
blackness; then the bluish green and green shade off into rapidly 
darkening gray, and finally the yellow and yellowish green lose 
their identity and merge into the night. At the same time the 
hues even of simple colors change, scarlet fading into orange, 
orange into yellow, and green into bluish green. 

" JT^Z\ 

I ire. > 

" till v it 

» £3 A - 

" -, 1 \ \ 

* 4-it % ^ 

I TV \ *v 

" ft K V 

M' ^s^- 

A ; : i l . I. 

Fig. 11. — Effect of Faint Light on Color. 

Obviously, complicated composite colors must vary widely under 
such circumstances, for as the light grows dimmer their various 
components do not fade in equal measure. Pinks, for instance, 
generally turn bluish gray at a certain stage of illumination, owing 
to the extinction of the red rays. In fact, in a dim light the 
normal eye is color-blind as regards red, and one can get a rather 
good idea of the sensations of the color-blind by studying a set 
of tinted wools or slips of paper in the late twilight. 

The similarity of the conditions is strikingly illustrated in Fig. 1 1, 
which shows in No. 1 the distribution of luminosity in the spec- 
trum of bright white light to the normal eye, and in No. 2 the 
luminosity of the same as seen by a red-blind eye. No. 3 shows 


the luminosity of the spectrum when reduced to a very small 
intensity and seen by the normal eye. The data are from Abney's 
experiments, and the intensity of No. 3 was such that the yellow 
component of the light corresponding to D of the spectrum was 
0.006 foot-candle. The ordihates of No. 2 and No. 3 have been 
multiplied by such numbers as would bring their respective maxima 
to equal the maximum of No. 1, as the purpose is to show their 
relative shapes only. The "red-blind" curve No. 2 shows very 
faint luminosity in the scarlet and orange and absence of sensa- 
tion in the crimson, while the maximum luminosity is in the 
greenish yellow. It is easy to see that the sensation of red is 
practically obliterated. 

But in No. 3 every trace of red is gone, and the maximum bril- 
liancy has moved up into the clear green of the spectrum at the 
line E. With a still further reduction of intensity, the spectrum 
would fade into gray as just noted, while a slight increase of light 
would cause No. 3 closely to approximate No. 2. 

Starting with the normal curve of luminosity No. 1, the peak 
of the curve being one candle power, the light at B would dis- 
appear if the illumination were reduced to 0.01 of its initial value, 
that at C at about 0.0011, at D 0.00005, at E 0.0000065, at 
F 0.000015, and at G 0.0003. 

Now the practical application of these facts is manifold. Not 
only do they explain the odd color effects at twilight and dawn, 
but it is worth noting that the cold greenish hue of moonlight on 
a clear night means simply the absence of the red and orange 
from one's perception of a very faint light; for dim moonlight 
is ordinarily not much brighter than would give curve No. 3. 
For the same reason a red light fades out of sight rather quickly, 
so that a signal of that color is not visible at a distance at 
which one of another. color and equal brightness would be easily 

Not only is the eye itself rather insensitive to red, but the 
luminosity of the red part of the spectrum of any light is rather 
weak, so that when the other rays are cut off by colored glass 
the effective light is greatly reduced. About 87 per cent of the 
effective luminosity of white light lies between the lines C (scarlet) 
and E (deep green), the relative luminosities at various points 
being about as follows: 


Line. Luminosity. 

B 3 

C 20 

D 98.5 

E 50 

b 35 

F 7 

G 0.6 

The luminosities of light transmitted through ordinary colored 
glasses of various colors is about as follows, following Abney's 
experiments, clear glass being 100: 

Color of Glass. Light Transmitted. 

Ruby 13.1 

Canary 82.0 

Bottle green 10.6 

Bright green (signal green No. 2) 19.4 

Bluish green (signal green No. 1) 6.9 

Cobalt blue 3.75 

These figures emphasize the need of a very powerful source, 
if it is necessary to get a really bright-colored light. It is worth 
noting that red in itself is a particularly bad color for danger 
signals on account of its low luminous effect, and were it not 
for the danger of changing a universal custom and the selective 
effects of atmospheric absorption, red should be the "clear " signal 
and green the danger signal, the latter color giving a much brighter 
light, and thus being on the average more easily visible. In fact, 
so-called red signal lights transmit the orange very freely, also 
the yellow, and even a little yellowish green, a pure deep red 
having so slight luminosity as to be quite impracticable. To 
obtain the necessary contrast the "green" signals usually verge 
upon blue-green to the detriment of their brilliancy. 

It is easy to see that any artificial illuminant is at a con- 
siderable disadvantage if at all strongly colored; for not only 
does a preponderance of red or green rays injure color percep- 
tion, but the luminosity of such rays is generally rather low, and 
they do not compensate for their presence by giving greatly in- 
creased illumination. 

Owing to this fact the effective illumination derived from ordi- 
nary sources of light is quite nearly proportional to the intensity 
of the yellow component of each. Crova has based on this rule 
an ingenious approximate method of comparing the total intensity 
of colored lights by comparing the intensities of their yellow rays, 


either from their respective spectra or by sifting out all but the 
yellow and closely adjacent rays by means of a colored screen. 

Certainly for practical purposes the rays at the ends of the 
spectrum are not very useful. So far as the ordinary work of 
illumination goes, white or yellowish-white light is desirable, and 
the only practical function of strongly colored lights is for signal- 
ing and scenic illumination. 

The general effect of strongly colored lights is to accentuate 
objects colored like the light and to change or dim all others. 
Lights merely tinted produce a similar effect in a less degree. 
Bluish and greenish tinges in the light give a cold, hard hue to 
most objects, and produce on the face an unnatural pallor; in 
fact, on the stage they are used to give in effect the pallor of 
approaching dissolution. Naturally enough such light is unfitted 
for domestic illumination, as, aside from its effect on persons, it 
makes a room look bare, chill, and unfurnished. In a less degree 
a similar effect is produced by moonlight, which, from a clear 
sky, is distinctly cold, the white light growing faintly greenish 
blue as its diminishing intensity causes the red to disappear. 

On the other hand, a yellow-orange tinge in the light seems to 
soften and brighten an interior, giving an effect generally warm 
and cheery. This result is extremely well seen in stage firelight 
effects. Strongly red light is, however, harsh and trying, so that 
it should generally be carefully avoided. 

While it is not easy to predict accurately the effect of tinted 
lights upon various delicate shades without a careful study of the 
light rays forming each, the average effects relating to the simpler 
colors are summarized in the following table. It is compiled from 
the experiments of the late M. Chevreul, for many years director 
of the dye works of the Gobelins tapestries. The colored lights 
were from sunlight sifted through colored glass, and the effects 
were upon fabrics dyed in plain, simple colors. 

The facts set forth in this table show well what should be 
avoided in colored illumination. As regards various shades of 
the same colors, it must be remembered that light shades are 
merely the full, deep ones diluted with white, which is itself 
affected by the color of the incident light. In a general way, 
therefore, one can use this table over a wider range than that 
written down. 

For instance, a very light red in blue light would look blue 



with a mere trace of violet, while in yellow light it would be 
bright yellow with a very slight orange cast. Generally, a very 
light color viewed by colored light will be between the effect pro- 
duced on the full color, and that produced by the light on a 
white surface. Similarly, a light faintly tinged with color will 
only slightly modify the tone of a colored object in the direction 
indicated for the full-colored light in the table. 

Color of 

Color of Light Falling upon Fabrics. 













Faint vio- 




Light yel- 





Intense red 





Red- violet 





Faint yel- 
low slight- 
ly green- 





Light red 

Yellow • 






tinged with 
faint red 














Bluish gray 






Vivid blue 







on orange 





blue- violet 







Dark blue- 

Deep blue- 












But delicate shades from modern dyestuffs, which often absorb 
the light in very erratic ways, as in Fig. 10, are a different matter, 
and do not obey any simple laws. On the other hand, pure 
colors, in the sense in which the scarlet around the C line of the 
spectrum is pure, act in a fashion rather different from that 
shown in the table, which pertains to standard dyestuffs which 
never are anywhere near being pure colors. However, as arti- 


ficial illumination has to do only with commercial pigments and 
dyes, the table serves as a useful guide in judging the effects 
produced on interior furnishings by change in the color of the 

Of common illuminants, none except for the mercury arc and 
the flame arc have any very decided color, yet most are somewhat 
noticeably tinged. One can tabulate them roughly as follows: 

IUuminant. Color. 

Sun (high in sky). White. 

Sun (near horizon). Orange-red. 

Skylight. Very bluish white. 

Electric arc (short). White. 

Electric arc (long). Bluish white to violet. 

Flame arc. Commonly, yellow. 

Mercury arc. Bluish green. 

Nernst lamp. Yellowish white. 

Tungsten lamp. .Yellowish white. 
Incandescent (normal), carbon. Yellowish. 
Incandescent (below voltage), Orange to orange-red. 


Acetylene flame. Yellowish white. 

Welsbach light. Yellowish to greenish white. 

Gaslight (Siemens burner). Whitish yellow tinge. 

Gaslight, ordinary. Yellowish to pale orange. 

Kerosene lamp. Yellowish to pale orange. 

Candle. Orange-yellow. 

Outside the earth's atmosphere the sun would look distinctly 
blue, while its light, after thorough absorption in the earth's 
atmosphere, gets the blue pretty completely sifted out, so that 
the light from the eclipsed moon, once refracted by the earth's at- 
mosphere and then reflected through it again, is in color a deep 
coppery red. 

Arc lights vary much in color, from clear white in short arcs 
with comparatively heavy current to bluish white or whitish violet 
in long arcs carrying rather small current. The modern inclosed 
arcs err in the latter direction, and give tolerable color effects only 
with yellowish white inner globes or shades. Incandescents, cs 
generally worked, verge upon the orange. Of the luminous flames 
in use, only acetylene comes anywhere near being white, although 
the powerful regenerative burners are a close second. Incandes- 
cent gas lamps, &t first showing nearly white with a very slight 
greenish cast, ordinarily acquire a greenish or yellowish-green tinge 
after burning for some time. 

It is evident, then, that a study of the color effects produced 


by colored illuminants is by no means irrelevant, for distinct tinges 
of color are the rule rather than the exception. 

But this is not at all the whole story, for the general color of 
the illumination in a given space depends not only on the hue of 
the illuminant, but upon the color of the surroundings. Colored 
shades, of course, are in common use; sometimes with a definite 
purpose, more often from a mistaken notion of prettiness. Used 
intelligently, as we shall presently see, they may prove very valu- 
able adjuncts in interior illumination. 

But far more important than shading is the modification in the 
color of the light which comes from selective reflection at surfaces 
upon which the light falls. In every inclosed space light is re- 
flected in one way or another from all the bounding surfaces, and 
at each reflection not only is the amount of light profoundly 
modified, but its color may undergo most striking changes. It is 
this phenomenon that gives its greatest interest to the study of 
color in illumination. Its importance is not always readily recog- 
nized, for few persons pay really close attention to the matter of 
colors, but now and then it obtrudes itself in a way that forces 

Take for example a display window lined with red cloth and 
brightly illuminated. Passing along the sidewalk, one's attention 
is immediately dr^wn to a red glow upon the street, while the 
lights themselves may be ordinary gas jets. To get at the signifi- 
cance of this matter, we must take up the effect of reflection and 
diffusion in modifying the amount and quality of light. 



To begin with, reflection is of two kinds — in their essence, the 
same, yet exhibiting very different sets of properties. The first, 
regular or specular reflection, may be best exemplified by the reflec- 
tion which a beam of light undergoes at the surface of a mirror. 
The beam strikes the surface and is reflected therefrom in form 
as sharp and as distinct as it was before its incidence, and in a 
perfectly definite direction. 



Fig. 12. — Regular Reflection. 

The character of this regular reflection is very clearly shown in 
Fig. 12. Here B is the reflecting surface — a plane, polished, bit 
of metal, for instance. AB is the incident ray and BC the re- 
flected ray. In such reflection two principal facts characterize the 
nature of the phenomenon. In the first place, if a perpendicular 
to the surface of the mirror — as BD — is erected at the point of 
incidence, the angle ABD is always precisely equal to the angle 
DBC. In other words, the angle of incidence is equal to the angle 
of reflection, which is the first law of regular reflection. Moreover, 
the incident ray AB, the normal to the surface at the point of in- 
cidence BD, and the reflected ray BC are all in the same plane. 



In this ordinary form of reflection, such as is familiar in mir- 
rors, the direction of the reflected ray is entirely determinate, 
and, in general, although the reflected ray has lost in intensity, 
it is not greatly changed in color. A polished copper surface, to 
be sure, shows a reddish reflection, and polished gold a distinctly 
yellowish reflection. Only in certain dyestuffs which exhibit a 
brilliant metallic reflection is the color strongly marked. In other 
words, a single reflection from a good, clean, specularly reflecting 
surface does not usually very greatly change either the intensity 
or the color of the reflected beam. The angle of incidence affects 
the brilliancy of the reflection somewhat, but the color only im- 
perceptibly. In the art of practical illumination regular reflec- 
tion comes into play only in a rather helpful way, and kindly 
refrains from complicating the situation with respect to color or 

The second sort of reflection is what is technically known as 
diffuse reflection. This term does not mean that the phenomenon 
itself is of a totally different kind from regular reflection, but, 
nevertheless, its results are totally different. No surface is alto- 
gether smooth. Even with the best polished metallic mirrors, 
while the reflected image is perfectly distinct at ordinary angles 
of reflection, it is apt to become slightly hazy at grazing inci- 
dence — that is, when the incident and reflected beams are nearly 
parallel to the surface. This simply means that under such con- 
ditions the infinitesimal roughness of the reflecting surface begins 
to be in evidence. 

To get an idea of the nature of diffuse reflection, examine 
Fig. 13. In this case a section of the reflecting surface is rough, 
showing grooves and points of every description — - in fact, nearly 
everything except a plane surface. Consider now the effect of a 
series of parallel incident beams — numbered in the figure from 
1 to 10 — falling upon the surface. Each one of them is reflected 
from its own point of incidence in a perfectly regular manner; 
yet the reflected rays, on account of the irregularity of the sur- 
face, lie in all sorts of directions and, moreover, in all sorts of 
planes, according to the particular way in which the surface at 
the point of incidence is distorted. Diffuse reflection, therefore, 
scatters the incident beam in all directions, for the roughnesses 
of an unpolished surface are generally totally devoid of any reg- 
ularity. The spot upon which a beam falls, therefore, radiates 


light in a diverging cone and behaves as if it were really 

Some consideration of the nature of this diffuse reflection will 
bring to light a fact which in itself seems rather surprising: namely, 
that the total intensities of the two kinds of reflection are not so 
different from each other as might appear probable at first thought 
— provided the roughness of the unpolished surface is not on too 
small a scale; for each of the incident rays in Fig. 13 is reflected 
from the surface just as in the case of Fig. 12, in a perfectly 
clean, definite, way, and there is no intrinsic reason why the 
intensity of this elementary ray should be any more diminished 
than in the case of regular reflection. 

Fig. 13. — Diffuse Reflection. 

A little inspection of Fig. 13, however, shows that rays Nos. 5 
and 10 are twice reflected before they get fairly clear of the sur- 
face, and if one went on drawing still more incident rays and fol- 
lowing out the figure on a still finer scale, a good many other rays 
would be found to be reflected two or more times before finally 
escaping from the surface. Such multiple reflection, of course, 
diminishes the intensity of the light just as in the multiple reflec- 
tion from mirrors; for there is always a little absorption, selec- 
tive or otherwise, at any reflecting surface. Thus, while the 
difference in the final intensities of light regularly and diffusely 
reflected is not so great as might be imagined, it still does exist, 
and for a perfectly logical reason. 


To go into the matter a little further — suppose the rough 
surface of Fig. 13 to be not heterogeneous, but made up of a 
series of grooves having cross sections like saw teeth. On exam- 
ining the reflection from such a surface we should find a rather 
remarkable state of affairs, for the course of reflection would 
then vary very greatly with the relation between the direction 
of the incident light and the surfaces of the grooves in the 
reflecting surface. 

Light coming in one direction, i.e., so as to strike the inclined 
surfaces of the grooves, would get clear of the surface at the 
first reflection, and the intensity of the reflected beam would 
have a very marked maximum in one particular direction. A 
beam falling on the reflecting surface in the other direction, 
however, — that is, on the perpendicular sides of the saw-tooth 
grooves, — would suffer several reflections before escaping from 
the grooves, and hence would lose in intensity, might be changed 
in color, and might be considerably diffused. This sort of phe- 
nomenon one may call asymmetric reflection. As we shall pres- 
ently see, it plays a somewhat important part in some very 
familiar phenomena. 

Reflection from ordinary smooth but not polished surfaces par- 
takes both of the nature of regular and diffuse reflection, and is, 
in fact, a mixture of the two phenomena, there being a general 
predominant direction of reflection plus a certain amount of diffuse 
reflection. This sort of thing is very commonly met with in prac- 
tical illumination. Fig. 14, from Trotter's experiments, shows the 
relative reflection at various angles of incidence from common 
Bristol board and from the matt surface of freshly set plaster of 
Paris and several other materials. The specular reflection in the 
first-named is very strong. Such a surface gives a glaring reflec- 
tion of artificial illuminants at certain angles, and the effect upon 
the eye is distressing. Trouble from this source is common in 
schools and in counting rooms, where the glare from too highly 
calendered paper has to be endured for long hours. 

The light from artificial illuminants usually falls on painted walls, 
on tinted papers with surfaces more or less regular, on fabrics, and 
on various rough or smooth objects in the vicinity. If these sur- 
rounding surfaces are colored, — as in the case discussed a little 
while ago, — some curious results may be produced. Of course, 
light reflected from a colored surface is colored, as we have seen 



already, but the manner in which it is colored is by no means 

When white light falls upon a matt colored surface, the reflection 
is generally highly selective as regards color. Fig. 15, from Abney's 
data, shows clearly enough the sort of thing which occurs. It 
exhibits the intensity of the reflected light in each part of the spec- 
trum when the reflecting surface is colored. The surfaces in this 
case were smooth layers of pigment. Curve No. 1 is the light 





S u 



O 12 


* 10 








10 20 


30 40 50 

Fig. 14 


80 70 80 00 

reflected from a surface painted cadmium-yellow; No. 2, Antwerp 
blue; No. 3, emerald green. Each curve shows a principal reflec- 
tion of the color of the pigment, reaching a rather high maximum 
value, but falling off rapidly in parts of the spectrum other than 
that to which the predominant pigment color belongs. As has 
been already shown, pigment colors are nearly always impure, and 
this fact is strikingly exhibited in the shape of the curves. The 
color of the main body of light reflected from any one of these 
surfaces is plain enough. 



The visible color of the light is, however, strongly influenced by 
the character of the surface. A shiny enamel paint, for example, 
will reflect specularly a good deal of light which is not strongly in- 
fluenced by the pigment, but is reflected from the surface of the 
medium without much selective action; consequently, there will be 
in the reflected light both light which has taken the color of the 
pigment and light unchanged in color. In other words, when 
viewed by reflected light, the pigment color is mixed with white, 
and when we have a perfectly simple pigment color — such as is 
not found in practice — this would lead merely to lightening the 
tint. It may, however, have results much more far-reaching; for 






1 ! 1 















■ — 

2 70- 


o 00- 

L J 

l > 


2 50- 

K 1 

1 1 



§ 40- 




































# Fig. 15. — Selective Reflection. 

an admixture of white ljght in sufficient quantity is able to shut 
out the distinct perception of any color, diluting it until it becomes 

The effects of this dilution are most marked in the ends of the 
spectrum — the brighter colors at the middle being least affected 
by the admixture of white light; hence the fact that such a surface 
as we have been considering, reflecting a mixture of white and 
colored light, may produce a change not only in tint, but in the hue 
of the color, if the color, as usual, is composite. For example, a 
purple in enamel paint might — according to its composition — 
look pinkish or light blue if the surface reflection of white light 
were particularly strong. If the pigmented surface is not shiny 


but capable of considerable reflection of colored light, another 
phenomenon may appear. 

Fig. 16 shows curve No. 3 of Fig. 15, emerald green pigment, 
and below it a similar curve, resulting from a second reflection 
of the light selectively reflected from a pigment of that color. 
Assuming what is nearly in accordance with the fact, — that the 
second reflection follows closely the properties of the first, — the 
result is obviously to intensify the green of the reflected light. 
The clear green portion of the light reflected from this particular 
pigment is practically embraced between the dotted lines P and 
Q of Fig. 16. After one reflection the area under the curve 

Pig. 16. — Effect of Multiple Reflection. 

embraced by these two lines is about 42 per cent of the whole. 
After two reflections it has risen to 55 per cent, and each succes- 
sive reflection — while greatly reducing the intensity of the re- 
flected light as a whole — will leave it greener and greener. 

Consequently in diffuse reflection those rays which are reflected 
several times before escaping from the surface are strongly colored, 
and the more such multiple reflections there are the more pro- 
nounced is the selective coloration due to reflection; hence, ordi- 
nary colored surfaces, from which diffuse reflection takes place, 
are apt to take very strongly the color of the pigment — more 
strongly, perhaps, than a casual inspection of the pigment would 


Now, as we shall presently see, in any inclosed space the light 
reflected from the bounding surfaces is a very considerable por- 
tion of the whole, and, therefore, if these surfaces are colored, 
the general illumination is strongly colored also, whatever the 
illuminant may be; in other words, colored surroundings will 
modify the color of the illumination just as definitely as a 
colored shade over the source of light. In planning the general 
color tone of a room to be illuminated, it must be remembered 
that if the walls are strongly colored the dominant tone of the 
illumination will be that of the walls rather than that of the 

An interesting corollary resulting from Fig. 16 sometimes appears 
in the colors of certain fabrics. If the surface fibers of the fabric 
lie in one general direction the light reflected from that fabric, 
which determines its visible color, follows somewhat the same 
laws laid down for asymmetric reflection, discussed in the case of 
Fig. 13. 

Light falling on the fabric from the direction toward which 
the surface fibers run does not escape without profuse multiple 
reflection, and hence takes strongly the color of the pigment. 
Light, however, falling on the fabric reversely to the direction 
of the fibers undergoes much less multiple reflection, and is likely 
to be mixed with a large amount of white light hardly affected 
by pigment at all; hence, the curious phenomenon of changeable 
color in fabrics — for instance, a fine purple from one direction 
of illumination and perhaps very light pink from another. 

If, in addition to the effects resulting from an admixture of 
white light in certain directions of incidence, one also has the 
curiously composite colors sometimes found in modern dyestuffs, 
the changeable color effects may be and often are very conspicu- 
ous; the more so, since in such colors, by multiple reflection, or 
— what amounts to the same thing — by more or less complete 
absorption of certain rays, the resultant color may be very pro- 
foundly changed. 

Absorbing media sometimes show these color changes very 
conspicuously; as, for example, chlorophyll, the green coloring 
matter of leaves, which in a weak solution is green, but of which 
a very strong solution of considerable thickness transmits only the 
dark-red rays. Similar characteristics pertain to many modern 
dyestuffs, and result, in connection with the composite reflection 



which has just been explained, in some very extraordinary and 
very beautiful effects. 

From what* has just been said about color reflection it is obvious 
enough that the loss in intensity in a reflected ray may be very 
considerable, even from a single regular reflection under quite 
favorable conditions. Many experiments have been made to 
find the absolute loss of intensity due to reflection. This abso- 
lute value of what is called the coefficient of reflection — that 
is to say, the ratio between the intensities of the reflected and 
incident light — varies very widely according to the condition of 
the reflecting surface. It also, in case the surfaces are not with- 
out selective reflection in respect to color, varies notably with the 
color of the incident light. 

The following table gives a collection of approximate results 
derived from various sources. 

Material. JfcSESL. 

Highly polished silver .93 

Mirrors silvered on back .85 

Polished gold .80 

Highly polished brass .75 

Highly polished copper .75 

Polished platinum .63 

Speculum metal .65 

Polished steel .60 

Burnished copper .50 

The losses in reflection are due to absorption and to a certain 
amount of diffuse reflection mixed with the regular reflection. The 
above figures are for light in the most intense part of the spectrum 
and for rather small angles of incidence. For large angles of 
incidence — 85 degrees and more — the intensity of the reflected 
beam is materially diminished, owing probably both to increase 
in absorption and to diffuse reflection. 

Mirrors silvered with amalgam on the back, and various bur- 
nished metals sometimes used for reflectors, belong near the bottom 
of the table just given. Silver is decidedly the best reflecting sur- 
face; under very favorable circumstances the coefficient of reflection 
of this metal is in excess of 0.90. A very little tarnishing of the 
surface results in increased absorption and diffusion and a still 
further reduction of the intensity of the reflected ray. The values 
of these coefficients show plainly the considerable losses which may 
be incurred in using reflectors in connection with artificial lighting. 


So far as general illumination is concerned, the light diffused at 
reflecting surfaces is not by any means lost, but that absorbed is 
totally useless. In the case of ordinary reflecting surfaces, one 
deals with a mixture of regular and diffuse reflection, and in 
practical illumination the latter is generally more important than 
the former, for it determines the amount of light which reaches 
the surface to be illuminated in ways other than direct radiation 
from the illuminant. 

Obviously, if one were reading a book in a room completely 
lined with mirrors, the effect of the illumination upon the page 
would be vastly greater than that received directly from the source 
of light itself. On the other hand, a room painted black through- 
out would give very little assistance from reflection, and the illu- 
mination upon the page would be practically little greater than 
that received directly from the lamp. Between these limits falls 
the condition of ordinary illumination in inclosed spaces. Gen- 
erally speaking, there is very material assistance from reflection 
at the bounding surfaces. The amount of such assistance depends 
directly upon the coefficient of diffuse reflection of the various 
surfaces concerned, varying with the color and texture of each. 

As has been already indicated, diffuse reflection is rough, hetero- 
geneous, regular reflection, more or less complicated, according to 
the texture of the reflecting surface, by multiple reflections in the 
surface before the ray finally escapes; and, therefore, the coefficients 
of diffuse reflection are not so widely different from those of direct 
reflection as might at first sight appear probable, so far at least 
as the total luminous effect is concerned. 

In certain kinds of diffuse reflection there is considerable loss 
from absorption as well as from multiple reflections. This is con- 
spicuously the case in the light reflected from fabrics, where there 
is not only reflection from the surface fibers, but where the rays 
before escaping are more than likely to have to traverse some of 
them. This is illustrated in a rather crude but typical way in 
Fig. 17, which gives a characteristic case of asymmetric reflection. 
We may suppose that the beam of light falls upon a surface of 
fabric having a well-marked nap. In the cut aa is the fabric sur- 
face composed of inclined fibers or bunches of fibers. These fibers, 
although colored, are more or less translucent and are not colored 
uniformly throughout their substance. Owing to their direction, 
rays 1, 2, and 3 get completely clear of the surface of the fabric 



by a single reflection. These rays are but slightly colored, be- 
cause of the comparatively feeble intensity of the coloration of 
the individual fibers, which have a strong tendency to reflect white 
light from the shiny surface. 

On the other hand, rays 4, 5, and 6, inclined from the other 
direction, are several times reflected before clearing the surface, 
and in emerging therefrom have to pass through the bunches of 
translucent fibers that form the nap. As a result these rays are 
strongly colored. The amount of white light is very small and 
the structure of the surface has produced a marked changeable 

In reality, of course, few rays actually escape on a single reflec- 
tion, and those striking almost in line with the direction of the 

Pig. 17. — Asymmetric Reflection from a Fabric. 

fibers, as 4, 5, and 6 in the figure, may be reflected many times, 
so that the actual effect is an exaggeration of that illustrated. 

Moreover, the material of the surface fibers exercises a con- 
siderable influence on the amount and character of the selective 
coloration. Silk is especially well adapted to show changeable 
color effects, since its fibers can be made to lie more uniformly 
in the same direction than the fibers of any other substance, and 
they are themselves naturally lustrous, so as to be capable even 
when strongly dyed of reflecting, particularly at large angles of 
incidence, a very considerable proportion of white light. Being 
thus lustrous, they form rather good reflecting surfaces, and hence 
the light entangled in their meshes can undergo a good many 
reflections without losing so much in intensity as to dull con- 
spicuously the resulting color effect; besides, silk takes dyes much 
more easily and permanently than other fibers, and hence can be 
made to acquire a very fine coloration. 


Wool takes dye less readily, and it is not so easy to give the 
surface fibers a definite direction. They are, however, quite 
transparent and lustrous enough to give fine, rich colors. Cotton, 
unless "mercerized," is much inferior to both silk and wool in 
these particulars; hence, the phenomena we have been investi- 
gating are seldom marked in cotton fabrics. 

In velvet, which is a very closely woven cut-pile fabric, the sur- 
face fibers forming the pile stand erect and very closely packed 
together. It is difficult, therefore, for light to undergo anything 
except a very complex reflection, and practically all the rays 
which come from the surface have penetrated into the pile and 
acquired a strong coloration. The white light reflected from the 
surface of the fibers hardly comes into play at all except at large 
angles of incidence, so that the result is a particularly strong, 
rich effect from the dyes, especially in silk velvet. 

Cotton velvet, with its more opaque fibers, seems duller, and, 
particularly if a little worn, reflects enough light from the surface 
of the pile to interfere with the purity and intensity of the color. 
Much of the richness in color of rough colored fabrics and sur- 
faces is due to the completeness of the multiple reflections on 
the dyed fibers, which produces an effect quite impossible to 
match with a smooth surface unless dyed with the most vivid 

In practical illumination one seldom deals with fabrics to any 
considerable extent, but almost always with papered or painted 
surfaces. These are generally rather smooth, except in the case 
of certain wall papers which have a silky finish. Smooth papers 
and paint give a very considerable amount of surface reflection 
of white light, in spite of the pigments with which they may be 
colored. The diffusion from them is very regular, except for 
this surface sheen, and may be exceedingly strong. When light 
from the radiant point falls on such a surface, it produces a very 
wide scattering of the rays, and an object indirectly illuminated 
therefore receives in the aggregate a large amount of light. 

A great many experiments have been tried to determine the 
amount of this diffuse reflection which becomes available for 
illumination. The general method has been to compare the 
light received directly from an illuminant with that received 
from the same illuminant by one reflection from a diffusing 


The following table gives an aggregation of the results obtained 
by several experimenters, mostly from colored papers. 

MitAriAl Coefficient of 

Matena1, Diffuae Reflection. 

White blotting paper 82 

White cartridge paper 80 

White cardboard 74 

Ordinary foolscap 70 

Chrome-yellow paper 62 

Cream paper 56 

Light-cream paint 52 

Light-orange paper 50 

Pale-green paint 45 

Plain deal (clean) 45 

Yellow wall paper t 40 

Yellow-painted wall (clean) 40 

Light-pink paper 36 

Yellow cardboard 30 

Light-blue cardboard 25 

Brown cardboard 20 

Plain deal (dirty) 20 

Yellow-painted wall (dirty) 20 

Light emerald-green paper 18 

Dark-brown paper 13 

Vermilion paper 12 

Blue-green paper 12 

Cobalt-blue paper 12 

Dark-green paper 05 

Maroon paper 05 

Black paper 05 

Deep-chocolate paper 04 

French ultramarine-blue paper 035 

Black cloth 012 

Black velvet 004 

At the head of the list stands white blotting paper, which is 
really a soft mass of lustrous white fibers. Its coefficient of 
reflection — 0.82 — is comparable with the coefficient of direct 
reflection from a mirror. 

White cartridge paper is a good second, and partakes of the 
same general characteristics. 

Of the colored papers only the yellows, and pink or green so light 
as to give a strong reflection of white light from the uncolored 
fibers, have coefficients of diffuse reflection of any considerable 
magnitude. Very light colors in general diffuse well owing to 
the uncolored component of the reflected light, but of those at 
all strongly colored only the yellows are conspicuously luminous. 

Of course, all of the papers when dirty diffuse much less effec- 
tively than when clean, and the rough papers, which have the 
highest coefficients of diffuse reflection, are particularly likely to 
become dirty. 


A smooth, clean, white board and white-painted surfaces gener- 
ally diffuse pretty well, but lose rapidly in effectiveness as they 
become soiled. Greens, reds, and browns, in all their varieties, 
have low coefficients, and it is worth noticing that deep ultra- 
marine blue diffuses even less effectively than black paper coated 
with lampblack, which has a diffusion of 0.05 as against 0.035 
for the blue. Black cloth, with a surface rough compared with the 
black paper, diffuses very much less light; while black velvet — 
of which the structure is, as just explained, particularly adapted 
to suppress light — has a coefficient of diffusion conspicuously less 
than any of the others. A little dust upon its surface, however, 
is capable of reflecting a good deal of light. 

These coefficients of diffusion have a very important bearing on 
the illumination of interiors. It is at once obvious that — except 
in the case of a white interior finish or a very pale shade of color — 
the illumination received by any object is not greatly strengthened 
by diffused light from the walls. All of the strong colors, par- 
ticularly if dark, cut down diffusion to a relatively small amount, 
although it is very difficult to suppress diffusion with anything 
like completeness. 

One of the standing difficulties in photometric work is to coat 
the walls of the photometer room with a substance so non-reflectr 
ing as not to interfere with the measurements. Even lampblack 
returns as diffused light one-twentieth of that thrown upon it, 
and painting with anything less lusterless than lampblack would 
increase the proportion of diffused light very considerably. Walls 
painted dead black, and auxiliary screens, also dead black, to 
cut off the diffused light still more, are the means generally taken 
to prevent the interference of reflected light with the accuracy of 
the photometric measurements. 

In the case of any diffusing surface, or any reflecting surface 
whatever, for that matter, a second reflection has, at least approxi- 
mately, the same coefficient of reflection as the first, so that for 
the two reflections the intensity of the beam that finally escapes 
is that of the incident beam multiplied by the square of the 
coefficient of diffusion, and so on for further reflections. 

Inasmuch as in any inclosed space there is considerable cross- 
reflection of diffused light, the difference in the total amount of 
illumination due to reflection is even more variable than would 
be indicated by the table of coefficients given; for while the 


amount of light twice diffused from white paper or paint would 
be very perceptible in the illumination, that twice diffused from 
paper of a dark color would be comparatively insignificant. 

The color of the walls, therefore, plays a most important part 
in practical illumination, for rooms with dark or strongly colored 
walls require a very much more liberal use of illuminants than 
those with white or lightly tinted walls. The difference is great 
enough to be a considerable factor in the economics of the 
question in cases where artistic considerations are not of prime 
importance. The nature and amount of the effect of the bound- 
ing surfaces on illumination will be discussed in connection with 
the general consideration of interior lighting. 



Considering the fact that the annual sum spent by civilized 
peoples for illuminants may be reckoned by hundreds of millions 
of dollars, it is somewhat extraordinary that methods of measur- 
ing light and standards by which it is to be reckoned have been, 
and for that matter still are, in so unsatisfactory a state. Until 
very recently it would be well within bounds to say that no 
commodity of similar total valuation has been so roughly and 
inaccurately measured as light. 

At the present time we are beginning to reach a somewhat more 
satisfactory standard of precision. The fundamental difficulty 
with the measurement of light is that it is a physiological rathep 
than a physical quantity and involves the uncertainties inherent 
in physiological measurements. One can measure out a kilowatt 
hour of electrical energy, or a thousand cubic feet of gas, or a 
gallon of kerosene, with a degree of precision good enough from 
the commercial standpoint; but to compute the light produced 
by any source through direct measurements thereof is altogether 
more difficult. 

Until as late as 1909, at which date an informal international 
convention made a single unit, the international candle, standard 
in France, Great Britain, and the United States, each country was 
a law unto itself in units of light and their applications. Three 
questions are involved in getting a measurement of light. First, it 
is necessary to have a standard of reference or primary standard 
giving an amount of luminous energy to which the light to be 
measured can be referred. Second, it is necessary to have a unit 
of light, that is, a conventional quantity of light which may or 
may not be equal to the concrete thing used as a primary standard, 
but which represents a definite quantity in terms of which other 
lights are stated. Finally, it is necessary to have at least a con- 
veniently uniform if not ideally precise procedure for the actual 
photometric work. 

Light in the last resort is the measure of the value of the illu- 



minants which one purchases, and consequently, as the aggregate 
amount of the purchases is very great, the importance of suitable 
standards and methods has long been recognized. 

Historically, the oldest photometrical standard is the Carcel 
lamp, for many years used in France, both as a concrete standard 
of luminous intensity and as the unit in terms of which commercial 
light was to be measured. The Carcel lamp, invented just at the 
beginning of the nineteenth centur^, is an argand burner with a 
wick and chimney of specified dimensions, consuming colza oil, 
which is fed up to a uniform level at the burner by a clockwork- 
driven pump placed in the base of the lamp. The wick, therefore, 
draws from oil maintained at a constant level under practically 
uniform conditions. The lamp is regulated to burn 42 grams of 
oil per hour with a permissible variation of 4 grams on either side 
of the normal. 

Offhand, from general experience with oil-burning argand lamps, 
one would say that the Carcel lamp would give but an indifferent 
approximation to uniformity and would be neither particularly 
reliable nor satisfactorily reproducible as a standard. In spite of 
its unpromising character, it is nevertheless true that in the hands 
of the French photometricians, who are used to it, it has given 
surprisingly good results, and it is a curious fact that the compari- 
sons of the Carcel with the Hefner lamp used in Germany are more 
consistent than the comparisons between any other pair of primary 
standards which have been used. The Carcel gives a light of fairly 
good yellowish hue in amount nearly 10 candle power. 

The next oldest standard used in recent times is the so-called 
parliamentary sperm candle legalized in 1860. This was a candle 
made of spermaceti, weighing 1200 grains avoirdupois, and burning 
at the rate of 120 grains per hour. The permissible variation in 
rate of burning is from 110 to 130 grains per hour, the luminous 
intensity being assumed to vary directly with the rate of burning. 
The normal diameter of the candle is 0.8 inch at the top and 0.9 
inch at the base, and the wick is required to be composed of three 
strands, each of 18 threads. 

This is the candle which is the commonest legal standard in the 
present statutes of this country, and was for many years the stand- 
ard in England. It was and is very unsatisfactory as a primary 
standard, seldom manufactured so as to be close to the specifica- 
tions, and remarkably sub j ect to accidental variations. With great 


care in using, it can probably be coddled to a precision of plus or 
minus 2 or 3 per cent, with variations twice that amount altogether 
too common. It is fortunately now discredited and obsolescent. 

The two most important primary standards in common use are 
the Harcourt 10-candle-power pentane standard and the amyl 
acetate lamp of von Hefner-Alteneek, generally known as the 
" Hefner." These two standards embody the correct principle of 
burning a definite chemical substance easy to obtain in compara- 
tively pure state, in lamps of dimensions so specified that they can 
be accurately reproduced, and under definitely specified conditions. 
Moreover, both have been studied for a period long enough to 
reveal their idiosyncrasies, and are in very wide use. 

The pentane standard is employed by the London Gas Referees 
and the National Physical Laboratory of England as the official 
standard, and is being considerably em- 
ployed by American gas companies. 
The Hefner lamp is in universal use as 
a standard throughout Germany and 
German speaking countries, and to a 
very considerable extent elsewhere. 

The pentane standard is essentially 
an argand gas burner fed by air satu- 
rated by pentane vapor. The lamp and 
some of its parts are shown in section 
in Fig. 18. The carburetor is a rec- 
tangular box containing baffle plates, 
around which the air has to pass to be- 
come saturated in going down to the 

The burner itself is of the argand 
form with a steatite ring containing the 
outlets, and is surmounted at a height of 
p. I8 47 millimeters by a brass chimney sur- 

rounded by an annular space, passing 
through which the air supplied to the burner is preheated. The 
flame is without a surrounding chimney, but is protected by a 
conical shield cut away to allow the flame to be visible. The 
normal height of the flame is 2} inches. 

The apparatus is obviously somewhat intricate, but when care- 
fully handled under closely regulated conditions gives a rather 


satisfactory degree of precision. It is probably good within one 
per cent on either side of the normal when carefully used, and 
duly corrected for barometric pressure, humidity, and CO* in the 
air. A strong point in favor of the pentane standard is the con- 
siderable light it gives, substantially 10 candle power, and the 
fact that the flame is fairly white. Some care has to be taken to 

Fig. 19. 

secure pentane of adequate purity, as, while it is a definite chemical 
compound which can be obtained absolutely pure, the common 
source of the pentane used is such as to render probable its con- 
tamination with small amounts of other hydrocarbons which may 
vary the illuminating power of the gas. 

The Hefner lamp is simpler and more easily reproducible than 
the pentane standard. It is shown in section in Fig. 19, in which 


the essential dimensions in millimeters are given. Here A is the 
body of the lamp, closed by the cap B which carries the working 
parts. These are essentially the wick tube C and the wick- 
adjusting mechanism which consists of two worms /, /i, meshing 
into gears e, ei, which carry the wick wheels w, w t . 

A rotating cap h carries a pillar upon which is mounted the 
optical flame gauge K, which is merely a sighting apparatus by 
which the flame can be adjusted to exactly the required height. 
The wick tube carries a snugly fitting, woven cotton wick, adjusted 
to the top of the wick tube by a cap gauge furnished for the 
purpose. The standard height of the flame is just 40 milli- 
meters. The fuel is pure amyl acetate. The wick chars very 
little while the lamp is in use, but it should be kept evenly 



Liters per Cutnc Moter 
Fig. 20. 

The intensity of the light varies with the flame height, the 
barometric pressure, and with the moisture and CO2 present in 
the air. The corrections due to these several causes have been 
carefully worked out and are shown graphically in Fig. 20. In 
this the ordinates are proportional intensities. Curve a exhibits 
the variation of intensity with the proportion of COj in the air. 
Curve b is the variation of intensity with humidity. Curve c, 
read by the upper scale of abscissa;, gives the variation with the 
flame height in millimeters. The barometric correction is very 
small, the intensity varying about 0.01 per cent per millimeter 
decrease of pressure, the normal being 760 millimeters. 

The lamp should l>o used in a well-ventilated room free from 
draughts, as the small flame is somewhat sensitive, and should 
be allowed to burn about half an hour before beginning meas- 


urements. The light given by the Hefner lamp as described is 
0.9 of an international candle; at a flame height of 45 millimeters, 
other things remaining the same, the lamp gives just one inter- 
national candle. 

The chief objections to the Hefner are its small luminous 
intensity and the strong reddish color of the flame, which intro- 
duces into comparisons made with it the difficulties of color pho- 
tometry to a somewhat undesirable degree. When carefully 
handled, it, like the pentane standard, is probably accurate to 
about one per cent, although in the case of both these lamps vari- 
ations of double this amount in case of different lamps operated 
by different people should not create surprise. 

The only other primary standard of light of any importance 
is Violle's platinum standard, thus far of very little importance 
as a concrete standard of reference, but of great significance as 
being indirectly the basis of the international candle. In 1881 
at the Paris Congress, Violle proposed, as a standard of light, 
that radiated from a square centimeter of melted platinum at 
its point of solidification. 

In its original form the scheme of operations required not less 
than one kilogram of molten platinum re-fused for each new 
observation, and the apparatus was very troublesome to work 
with. Later modifications of the apparatus by Siemens, Petavel, 
and others have proved somewhat easier to operate, but the consen- 
sus of opinion is that as a primary standard it is more troublesome 
and less precise than either the Hefner form or the pentane lamp. 

Nevertheless the Violle standard was adopted by the Paris 
Electrical Congress of 1881, and the twentieth part of this unit, 
determined in practice by comparison with the Carcel lamp, has 
been considerably used in France under the name of the bougie 
decimale, which in turn has been adopted as the present inter- 
national candle. 

The international candle, therefore, is not a primary standard 
at all, but a unit of luminous value derived from intercomparison 
of the Carcel, Hefner, and pentane primary standards. By 1907 
it had become evident that there were outstanding differences 
between the relative values of the units of light commonly re- 
ceived, sufficient to demand commercial attention. The initiative 
in the matter was taken by the Illuminating Engineering Society, 
which appointed a committee on units and standards including 



distinguished foreign members of the society in France, England, 
and Germany, and charged it with the work of undertaking to 
obtain an international convention on a working unit of light. 

The work was actively taken up, in cooperation with the society, 
by the American Institute of Electrical Engineers, the American 
Gas Institute, and the National Laboratories of France, England, 
and the United States. 

As the result of elaborate intercomparisons between the primary 
standards of light in use, both directly and via incandescent lamp 
standards, at the three laboratories mentioned and at the Reich- 
anstalt, it was finally determined that the unit of light should be 
taken at the value of the bougie decimate, of which the Hefner 
standard should be taken as nine-tenths. 

The necessary concessions were made by the bodies interested 
to bring the values of commercial standards into harmony with 
this determination, and since June 1, 1909, this international 
candle has been the standard in France and in English-speaking 
countries, and is gradually winning adherence among other nations. 
The German practice still retains the Hefner as unit as well as 
primary standard, since its difference from the international candle 
is so considerable as to cause more or less commercial incon- 
venience. The incandescent lamp has been taken as the custodian 
of this unit value, since intercomparisons of incandescent lamps 
can be made with a relatively very high degree of precision; yet 
it must be remembered that flame standards, whether primary 
or secondary, have a direct and great value in determining the 
value of commercial name illuminants, inasmuch as by the use 
of flame standards the corrections for moisture, CO2, and baro- 
metric pressure become either negligible or so small as to be very 
readily made, while comparisons between flames and incandescent 
lamps involve all these corrections at their full values. The rela- 
tions between the intensities of the various primary standards 
and the international candle unit are given in the following table: 




International candle. 

Carcel lamp 


Old English candle.. 








Old English 



The secondary standard most in use, as just stated, is the in- 
candescent electric lamp. After a lamp has been aged by burning 
about 200 hours its candle power falls off only very gradually with 
further use for about an equal period, so that if a lamp stand- 
ardized after aging is used only for calibrating a working standard 
it remains reliable far within the errors of observation for a con- 
siderable period. Practical standards of a high degree of pre- 
cision are therefore readily available in the form of incandescent 

As a secondary flame standard the Methven screen has been 
considerably used, particularly in England. This standard is a 
powerful argand gas burner fitted with a chimney and having 
adjusted in front of it 1 J inches from the axis of the flame a 
blackened metal plate having a slot just in front of the flame 1 
inch high by 0.233 inch in width. A section of flame thus cut 
out is a very convenient and steady secondary standard of about 
2 candle power. In this country the Elliot lamp has proved a 
valuable adjunct in gas testing. It is a kerosene lamp, of the 
"student lamp" type, of which a definite area of the flame is 
exposed by a slot, as in the Methven screen. The section of the 
flat flame thus cut out remains sensibly uniform in light for a 
considerable period. These secondary standards and the 10-c.p. 
pentane primary standard are now rapidly displacing the discredited 
"standard " candle for gas testing, so that for all illuminants there 
is now a definite basis of reference to the international candle and 
to the Hefner, which is a determinate fraction thereof. 

Granted a unit of light intensity such as is furnished by the 
international candle, and a concrete standard representing it or 
a known multiple thereof, the fundamental process of light meas- 
uring is the determination of the intensity of some working source 
of light in terms of the unit. 

Photometry is the art of comparing light intensities, and as 
such it is the basis of the quantitative part of illuminating engi- 
neering. In the last resort, a photometric comparison depends 
upon the ability of the eye to detect small differences in lumi- 
nosity, and the photometer ordinarily is an instrument designed 
to present to the eye two similar juxtaposed surfaces, one lighted 
by the source of known intensity and the other by the source of 
unknown intensity, together with means of varying either or both 
of these intensities in a determinate manner. 


The power of the eye to recognize the minute differences of 
luminosity necessary to precise photometric measurement depends 
on the value of Fechner's fraction, to which reference has been 
made, and the fine art of photometer design is to so arrange the 
apparatus as to aid the shade perception of the eye in the most 
efficient possible manner. 

The earliest form of photometer, that of Bouguer, now more 
than a century and a half old, is absolutely typical of photo- 
metric principles. In diagram it is shown in Fig. 21. Here ab 

is a screen with an opaque partition 
L * ,a cd perpendicular to its middle point c. 

This screen may be a diffusing surface 
like cardboard, or a translucent ma- 
terial like thin paper or milky glass. 
LX - 'a j n t ne first case it is viewed from 

lg * the front, in the latter from the rear; 

L and U are the lights to be compared, the former illuminating 
the screen over ac y and the latter over be. When one of the 
lights is moved to or from the screen until the two halves of the 
screen are of equal apparent brightness, then the intensities of 
the lights are proportional to the squares of their respective dis- 
tances, d and d', from the screen. Obviously the dimensions of 
the screen should be small compared with d or d', so that the illu- 
mination may be sensibly uniform over each half; and these must 
be equal in reflecting or transmitting quality, so as to introduce 
no constant error due to dissimilarity of the two halves. Two 
areas thus merely juxtaposed with a black line of demarcation 
between them do not present the most favorable conditions for 
delicate, shade perception, and the betterment of these conditions 
has been the object of such improvements as have been made in 
the photometer. 

Of practical forms of photometer there are many, differing 
chiefly in the nature and arrangement of the luminous areas to 
be compared and in various details of convenience. The only 
radical departure from Bouguer's principles is in the case of the 
so-called flicker photometers, to be described presently. 

The two typical forms of photometer in most general use are 
the Bunsen and the Lummer-Brodhun, so called for the respective 
inventors of the comparison screens. The Bunsen photometer 
consists essentially of a graduated bar, with one of the lights to 


be compared at each end, merely for the purpose of enabling the 
distances of the two lights from the observing screen to be easily 
determined. On the bar slides a sight box containing the BunBen 
screen and carrying a pointer moving over the scale. The general 
disposition of the parts is shown in Fig. 22. 

Fig. 22. 

The length of the graduated bar is commonly 100 inches if 
graduated in English measure, 2.5 or 3 meters when in metric 
measure. The sight box as commonly made is shown in plan in 
Fig. 23. The Bunsen screen forms the middle partition shown in 
the box, upon which the light from the sources on either side 
falls. Two mirrors, mi and m*, placed substantially as shown, 

Fig. 23. 

enable one looking into the wide sight tube T to sec both sides 
of the screen at once. 

The Bunsen screen itself in its commonest form consists of a disk 
of opaque, matt-surfaced white paper with a sharply defined central 
spot made translucent by grease, usually paraffin. As already ex- 
plained, such a spot appears bright or dark on the general surface 
of the disk according as the illumination behind it is stronger or 
weaker than that on the front. With a screen in perfect condition 



the spot will nearly or quite disappear when the illuminations are 
equal. When the screen is so placed on the photometer bar that 
this condition is fulfilled, the intensities of the lights to be com- 
pared are respectively as the squares of their distances from the 
screen, this being the condition for equality of illumination, or in 
the more general case this same condition holds when there is 
equal contrast between spot and surface as seen in the two mirrors. 
To eliminate inequalities of appearance due to difference be- 
tween the two sides of the screen or between the two mirrors, the 
sight box is commonly made rotatable through an angle of 180 
degrees, so as to reverse the position of the screen and mirrors 
with respect to the two lights under comparison. These two lights 

Fig. 24. 

must be screened off from the observer so that they will not inter- 
fere with his judgment, and black screens with large central holes 
are commonly mounted on the bar so as to intercept stray light; 
and for the same reason it is desirable to operate in a darkened 
room with black walls. If the intercepting screens do their work 
very completely, the black wall surface is not absolutely necessary, 
but it is on the whole to be preferred. 

A very material improvement over the Bunsen grease-spot disk 
is the Leeson disk, which consists of a piece of thin white trans- 
lucent paper sandwiched between two pieces of opaque white paper 
with central star-shaped openings, so that the disk presents a 
sharply defined star, preferably with 10 or 12 narrow points, the 
whole being 1 inch to 2 inches in diameter. This Leeson disk is 
worked precisely like the Bunsen disk, but since its outlines are 
usually sharper the Leeson disk gives rather more precise settings. 


It is desirable, for accurate estimate either of equality of bright- 
ness or of contrast, that there should be no debatable ground 
between the two areas compared; that is, that the two should 
come sharply up to each other without a perceptible dark line 
between them. This condition is fulfilled more perfectly by the 
Lummer-Brodhun screen than by any other device yet contrived, 
and for precise laboratory work this is the screen usually adopted, 
the mounting of the other parts of the photometer being practically 
as in the Bunsen. The Lummer-Brodhun sight box complete is 
shown in plan in Fig. 24. The box is mounted on the photometer 
bar so as to be rotatable through 180 degrees, with its axis of rota- 
tion, uz> perpendicular to the bar. The screen proper, c>c\ d,d', is 

Fig. 26. 

a disk, usually of compressed magnesia, which gives a very perfect 
matt surface, upon the two sides of which fall normally the rays 
from the lights under comparison. This screen is viewed simul- 
taneously from both sides by means of the mirrors /i,/ 2 , and the 
right-angled prisms, A,B. A cross screen, x, serves to cut off 
scattered light. 

The prisms with the paths of the rays through them are shown 
in Fig. 25. The hypotenuse faces of the prisms are ground optically 
flat and clamped together. But, prior to clamping, the surface of 
A is recessed by sandblasting in vertical strips, ri,r a . When the 
prisms are clamped into optical contact, light falling on the hy- 
potenuse surface of B opposite these recesses is totally reflected, 
while in the intervening spaces, tifa, it is transmitted. The path 
of the rays is plainly shown by the dotted lines, and the result is 


that the odd-numbered rays received from c,c' via /i enter the 
sight field only through the contact faces h,fc, while the even- 
numbered rays from d,d' via/2 enter only by total reflection at ri,r 2 . 

The result is a sight field that looks like Fig. 26, each half-circle 
receiving light from one side of the screen and having superimposed 
upon it a trapezoidal area received from the other side of the screen. 
The sight box can be set for disappearance of these areas so as to 
present a uniform field, or by inserting two slips of glass mc } gb, 
Fig. 25, the trapezoidal areas can be slightly darkened so that 
when everything is in balance one sees two shaded areas in a uni- 
form field and thus works by equality of contrast. The field is 
viewed by a magnifying lens w, Fig. 24, set in a sliding eye-tube, 0. 

This arrangement of sight field is wonderfully sensitive in 
showing small variations of intensity, certainly a good deal 
more sensitive than the Bunsen or Leeson disks, especially when 
the lights compared are of similar color. Opinions differ among 
photometrists as to their relative sensitiveness, but the general 
result of experience seems to be that while the mean error of a 
single setting with the Bunsen or Leeson disk is likely to be nearly 
1 per cent, that obtained with the Lummer-Brodhun screen worked 
for contrast will be less than one-half per cent, and under favorable 
circumstances down to one-third or one-fourth per cent. Most 
operators prefer the Bunsen or Leeson disk for lights differing 
materially in color, as, for example, in the comparison of a carbon- 
filament incandescent lamp with a tungsten lamp or with incan- 
descent gas. Under such circumstances photometric settings are 
likely to show nearly double the mean errors just referred to, the 
Lummer-Brodhun suffering relatively somewhat more than the 

The difficulty of forming a just estimate of equality of bright- 
ness or of contrast in the case of two illuminations differing in 
color is so great as to constitute the largest outstanding source 
of error in photometry. It is not putting the case too strongly 
to say that there is no simple method of comparing lights differing 
much in color with a reasonable degree of precision. Various 
subterfuges have been adopted for such comparisons, which are 
convenient rather than satisfactory. Perhaps the best of them 
is the use of the so-called " flicker " photometer. 

The essential point of this instrument is the rapid exchange of 
the two illuminations to be measured with respect to the sides 




^--Y^p -J/i 

of the viewing screen, and the principle may be by one arrange- 
ment or another applied to almost any kind of screen. One of 
the best known and simplest forms of flicker photometer is the 

The essential part of this device is a disk of plaster of Paris, 
say five-eighths inch thick and 2| inches in diameter, rotating on 
an axis parallel to the photometer bar. The disk is molded so as 
to form a species of double conoid, which, when looked at edge- 
wise, presents a surface illuminated by the two sources in turn, 
the transition from one reflecting position to the other being 
gradually accomplished by the rotation of the disk. Fig. 27 shows 
three edgewise views of the disk, 
which, when looked at from above 
as indicated, gives an inclined 
surface to the right, a wedge 
reflecting from left and right, 
and an inclined surface to the 
left in succession as the disk 
turns. The disk is mounted in a 
box, with apertures on the sides 
to admit light, a viewing tele- 
scope, and a spring or other motor 
for rotation with means for regu- 
lating the speed. 

Obviously, if one of the lights is noticeably brighter than the 
other and the disk is turning at a moderate rate of speed, a strong 
flickering sensation will be produced, which disappears when the 
illumination on the two sides of the disk becomes equal. In 
the comparison of colored lights the blending of the two colors 
by the rotation of the disk diminishes very considerably the 
troublesome contrast presented by two juxtaposed colored fields, 
the colors being, as it were, optically averaged and the screen 
being then adjusted until the flicker disappears. 

The flicker instruments generally give, in the comparison of two 
lights differing considerably in color, as, for example, a very low 
efficiency carbon incandescent and a tungsten incandescent, results 
differing by several per cent from those obtained from the same 
lights compared by ordinary photometer screens. In the case of 
lights varying still more in color, the difference may rise to 10 or 
even 20 per cent. The weight of the evidence indicates that the 

Fig. 27. 


flicker principle is of much value in comparing lights of different 
colors, since the readings obtained by a number of observers on the 
same pair of lights generally show for smaller differences with the 
flicker photometer than with the ordinary form. 

Yet even this consistency is not wholly satisfactory evidence of 
precision. Generally speaking, the flicker instruments do not allow 
of as quick and easy balance as those with fixed screens, although 
the precision of the balance when obtained is much better than 
seems probable while it is being made. Very high precision is 
claimed for some of these instruments, particularly by their inven- 
tors, but most photometrists hardly expect to obtain with them 
any smaller average deviations than with the ordinary grease 
spot. Flicker instruments work badly in weak illumination. 

Two other schemes for avoiding the color difference in the com- 
parison of lights are worth noticing here. One of them is the use 
of standard color screens, the coefficient of absorption of which 
can be obtained by the spectrophotometer and which serves to 
reduce the two lights under comparison to approximately the same 
color. The other is the preparation of a set of secondary standards 
presenting only slight successive differences of color and yet on the 
whole reaching a wide difference. This divides the color-matching 
difficulty into steps, as it were, which renders photometric settings 
much easier. Both these methods are somewhat roundabout and, 
like the flicker photometer, lessen rather than abolish the color 
difficulty. Probably any one of them yields results sufficiently 
good for most commercial purposes, but where scientific precision 
is required all leave much to be desired. 

For a full discussion of the problem of 'heterochromatic photom- 
etry the reader is advised to consult the various treatises and 
papers dealing specifically with this matter, which will be found in- 
teresting and instructive, although not always convincing. Photom- 
etry is at best a process involving physiological and psychological 
quantities of a somewhat indeterminate character. Even with the 
same instrument and comparing lights of similar magnitude and 
color, different observers are likely to find slightly different ratios, 
and the same is true if one considers even the observations of the 
same observer on different days. These differences may amount 
to a considerable fraction of a per cent, plus or minus. 

The fundamental thing in photometry is to make the best use of 
the instrument at hand, to hold the lights under comparison abso- 


lutely steady, and to eliminate as far as possible constant errors 
due to the apparatus. In the comparison of flames, this implies 
very careful sheltering from draughts and- close attention to the 
conditions of the air and of the standard flames. In comparing 
incandescent lamps with each other, the utmost care must be taken 
to hold the yoltages uniform during the comparisons. It is best 
to put the lamps upon the same circuit, preferably supplied by a 
storage battery or by a special generator, and to arrange the 
connections somewhat as shown in d 

Fig. 28. The lamp sockets A and B 
should be connected at opposite ends 
of the photometer bar by heavy cop- 
per leads, to which can be joined, by 
closing a switch, the circuit terminals, 
a and d. In circuit with a or d should 

be a resistance of large capacity, R, " ' — VWVV 

for approximate adjustment of the R 

voltage. R should be able to carry g ' 

the maximum current likely to be needed without sensible change 
of resistance, and should be either in very numerous steps or con- 
tinuously adjustable like a water rheostat. Each lamp socket, A, 
B (Fig. 28), should also be in series with its own adjusting rheostat, 
ry. These rheo&tats should have a capacity sufficient to give a 
range of say 10 to 20 volts at the lamps, and the less their tem- 
perature variations and the finer their gradations the better. The 
voltmeter, E, should be a double-scale instrument of extra-high 
resistance, with fine graduations, and preferably with illuminated 
scale. A high scale to 250 volts and a low scale to 25 volts is a 
convenient arrangement. 

Switches should be provided to make the following voltmeter 
connections: (1), 6 to c on low scale; (2), a to d on high scale; (3), 
b to d on high scale; (4), c to d on high scale. Position 1 is the 
differential connection used in holding the lamps at voltage; the 
others are for the approximate adjustments. In comparing lamps 
of very different voltage, it is convenient to connect be through 
the high scale. A reversing switch should be in be unless the volt- 
meter scale reads both ways. Ammeters or wattmeters when used 
should be connected between b and B f or c and A, so as to get 
inside the voltmeter, the current capacity of which ceases to be 
negligible when one tests lamps of small candle power or very 



high efficiency. All rheostat handles and switches should be within 
easy reach of the observer at the voltmeter. 

The method of working which should be employed whenever it 
is possible is that of substitution. At one end of the photometer 
bar, as at A, should be placed a properly aged lamp to be used as 
a working standard. The primary or tested standard is generally 
at B and its relation to A is determined without reversal of the 
photometer screen. Then without any changes in the apparatus 
B should be replaced in succession by the lamps to be tested and 
photometric balance secured. By this means any want of sym- 








^* ••«• 








■— - 









/ / 


• 1<M 













^ ' 





92 96 100 105 

£ Normal Volts 

108 113 


Fig. 29. 

metry in the screen or in the photometric apparatus generally is 
eliminated, since the ratios of B to A, and of the unknown lamps 
to A , are determined under exactly the same conditions. The one 
important matter during the test is to hold the voltage constant, 
for while the lamps may be rigorously held at the proper difference 
of voltage it does not follow that the light ratios will be the same 
when the absolute voltages vary, since different lamps gain or lose 
light to different extents with change of voltage, and consequently 
if the voltage on the testing circuit varies an error may be intro- 
duced. The nature of this error is shown in Fig. 29, which dis- 
plays variations in candle power produced by change of voltage 


on ordinary carbon, G.E.M., tantalum, and tungsten lamps. It 
will readily be seen that a variation in the general voltage of even 
a half a volt is sufficient to produce measurable errors even when 
the lamps are held rigorously at the required difference of voltage 
for which they are supposed to be normal. It is easy to hold the 
circuit voltage close enough for comparison of different carbon 
lamps or of different metallic filament lamps, but in comparing a 
carbon with a metallic filament lamp far greater care is necessary. 
It is advisable, therefore, to have working standards of both kinds. 
In comparing electric with other lamps, it is obvious that the volt- 
age control must be of the most exact kind in order to avoid 
material errors. 

In the simplest form of comparison between two lights the com- 
parison is made in a single azimuth only. Standard lamps have 
commonly a certain marked direction for which they are standard- 
ized, and to this direction they must be exactly oriented. It has 
been rather common practice, inasmuch as incandescent lamps are 
often rated by their mean horizontal candle power, to rotate the 
lamps during measurements, and indeed some standards are made 
for use during rotation. To this end a socket capable of revolu- 
tion by a motor at three or four times a second is often provided 
at one end of the photometer bar, as at B, and in this the 
standards and the lamps to be tested are to be used; the working 
standard at A, being merely a light of constant value, remains at 

In using these rotating sockets considerable care has to be 
exercised to prevent any variable drop in voltage at the contacts, 
and the tendency at present is to use the rotator less and less. 
Most modern incandescents, especially metallic filament lamps, 
give a nearly uniform horizontal distribution, and metallic fila- 
ment lamps in particular can not be rapidly rotated without risk 
of distorting the filament. Except in lamps with very peculiarly 
placed filaments, three measurements taken 120 degrees apart 
around the vertical axis will give the mean horizontal intensity 
within the limits of ordinary commercial accuracy. 

There are many devices for measuring the mean horizontal or 
the mean spherical intensities of light sources, most of them rather 
intricate and troublesome. Descriptions of these methods may be 
found in modern treatises on photometry, but elaborate explana- 
tion of them is out of place here. The one most satisfactory and 


convenient method of getting the mean spherical candle power, 
which is, in modern practice, the only suitable method of com- 
paring illuminants, is found in the sphere photometer commonly 
known as the Ulbricht sphere. The fundamental principle of this 
is that if a luminous source be placed at the center of a hollow 
sphere, the inner surface of which is a good diffuse reflector, the 
illumination at a peephole in the sphere shielded from direct rays 
received from the source is directly proportional to the mean 
spherical intensity of the source. A good account of the theory 
of this sphere is given by Dr. Bloch in the Elektrotechnische Zeit- 
8chrift of Jan. 18, 1906, page 63. The sphere itself is commonly 
made of heavy sheet metal, made up in two hemispheres separable 
by handles, or in large spheres by fitting them on a sliding track. 
The diameter depends on the purposes for which the photometer 
is intended; if for incandescent lamps, a meter is sufficient, while 
for arc lamps it should be not less than two meters. 

The interior of the sphere is usually painted dead white with 
a barium-sulphate paint. The light at the peephole may be bal- 
anced by any suitable photometer screen against a standard lamp, 
and the whole affair is calibrated by suspending in the sphere an 
incandescent lamp of which the mean spherical candle power has 
already been determined by the point-to-point method, that is, 
by measuring the intensity in a considerable number of directions 
uniformly spaced on the surface of an imaginary sphere surround- 
ing the lamp. A description of this point-to-point method may 
be found in the Journal of the Franklin Institute, September, 
1885, Supplement; and a valuable paper on the precautions to be 
taken in using the integrating sphere, by Sharp and Millar, is 
to be found in the Transactions of the Illuminating Engineering 
Society, Vol. Ill, page 502. For the testing of arc lamps, always 
a difficult matter, the use of the integrating sphere is almost a 
necessity, the only effective substitute for it being the integrating 
photometer of Matthews, which, though giving satisfactory pre- 
cision, is far more intricate. A description of this apparatus, 
important on account of the numerous and valuable measure- 
ments secured through its use, will be found in Transactions of 
the American Institute of Electrical Engineers, Vol. XX, page 69. 

From the standpoint of the illuminating engineer the most 
interesting photometers are those intended for portable use, by 
which one is enabled to measure intensity of light sources in situ, 


and the illumination received from them. The problems of meas- 
uring the intensity of a light and the illumination produced by it 
on a surface are virtually the same, since in either case the proc- 
ess of photometry is the balancing against a surface illuminated 
by a known light, another surface illuminated by the unknown 
light, and it makes no difference whether this second surface of 
the photometer screen receives its light from a lamp directly or 
from a surface illuminated by that lamp. The only essential 
difference is that the illumination photometer is calibrated with 
reference to the secondary source of illumination which is to be 
employed with it, that is, with respect to the surface either dif- 
fusely reflecting or diffusely transmitting light from the source 
under investigation. 

Fig. 30. 

A photometer for portable use is, however, a much more 
troublesome affair than one set up in the laboratory. The same 
photometer screens can be and are used in each, but the difficulty 
is in providing a reliable standard light and suitable means for 
modifying its intensity to secure a balance against the light to be 
measured without being driven into inconvenient complications 
or apparatus too bulky to be portable. The type of portable 
photometers is that of Dr. L. Weber. A general view of one of 
the forms of this instrument is shown in Fig. 30, and a diagram- 
matic section of it in Fig. 31. The instrument is composed of 
two main tubes, A, B, connected by a collar C, permitting one 



tube to rotate with respect to the other. A carries at its outer 
end a lamp case G, containing the standard light: in the earlier 
instruments a small benzine lamp, as shown in Fig. 30, in recent 
instruments more often an incandescent lamp operated from a 
storage battery. A screen of translucent glass, F, is fitted to slide 
back and forth along the tube A } being moved by the handle a, 
which carries a pointer over a scale on the side of the tube. The 
end of the other tube, B, is closed by a screen, G', also of diffusing 
translucent glass. At D is a Lummer-Brodhun prism viewed by 
the eyepiece E 7 usually provided with a right-angle prism to make 
observation convenient when the tube is in a vertical position. 


i — - 









Fig. 31. 

In use, the sight tube B is pointed at the source to be measured, 
with the screen G f in place, and a balance is then effected by 
moving F. Knowing, then, the distance, l if of the source, Ji, 
examined, and the intensity, 7, and distance, I, of the standard, 

7 2 

assuming the two diffusing screens to be exactly similar, or 


I 2 

h = & ~J2 I> 

wherein K is the ascertained ratio of the screens. Several diffus- 
ing screens are supplied with the instrument to increase its work- 


ing range, also a diffusing screen to act as a secondary source 
of illumination by reflection, G' being in this case removed. In 
practice it is found that the law of squares holds only very roughly 
for the positions of the movable screen F y so that for careful work 
the scale must be calibrated by experimental settings on lights of 
knoWBT intensity and distance. When such calibration is properly 
made in the laboratory, the instrument is capable of excellent 
work. When used for determining illuminations directly, the 
tube is left vertical and the light falls on the diffusing screen (?', 
or, this screen being removed, the sight tube is pointed at a white 
diffusing surface set at an angle with it. 

When properly calibrated for known illuminations falling on 
either of these screens, the Weber photometer works well for 
illumination measurements. In using it in this way the condition 
of the diffusing surfaces requires very close attention, and the 
calibration of this or any other instrument for a similar purpose 
requires to be very carefully watched, for it cannot be predeter- 
mined from the dimensions of the instrument and is subject to 
change without notice from the effects of dirt. 

A considerable number of portable photometers based on the 
general scheme of the Weber instrument have been devised and 
are in successful use. They differ chiefly in the means taken to 
vary the standard light in securing a balance. A fixed diffusing 
screen, the area of which is cut down by a cat's-eye; moving the 
lamp itself, varying the current or voltage applied to the lamp, 
are some of the methods employed, each of them in several different 
instruments. They are all effective provided close watch is kept 
on the calibration of the instrument, and not otherwise. The 
principal differences between these modified Weber photometers 
are differences of detail bearing on convenience of manipulation, 
which in most cases leaves much to be desired. 

As a class most portable photometers are only moderately 
portable; very few of them can be operated without the coopera- 
tion of two or more observers. Yet in skillful hands with proper 
calibration they all are capable of giving fairly good precision, — 
good enough at least for the conditions of their use. It must not 
be forgotten that portable photometers are not used to measure 
lights operated as they would be in the laboratory. On the 
contrary, they are commonly employed for the photometry of 
arc lights, the intensity of which is subject to accidental and 



periodical variations of 10 to 50 per cent; or for measuring incan- 
descent electric or gas street lights; the voltage and current being 
only approximately known in the former case and the pressure 
in the latter being very uncertain. 

When used for illumination measurements they are usually 
evaluating the effect of lamps, operated at unknown voltage or 
gas pressure and in unknown stages of deterioration^ in lighting 
interiors in which the conditions of wall reflection are unknown 
and subject to variations, and where the fittings cause local changes 
in illumination many times greater than the largest possible error 
of photometric balance. For work in the laboratory the portable 
photometers of the better class are capable of as good precision 
as fixed photometers, and the usual increase in the errors of 
measurement is due to the conditions of use rather than to intrinsic 
faults in the instruments. 

Fig. 32. 

A few instruments of a totally different class are in use for the 
estimation of illumination. These are photometers based on visual 
acuity; in other words, instruments depending upon the capacity 
of the eye for reading type in a dim light. One familiar type is 
the illuminometer of Houston and Kehhelly. It is essentially an 
extinction photometer, the light received from the source under ex- 
amination being varied until certain test characters cease to be 
visible. This illuminometer is shown in section in Fig. 32. Here 
X y X is a blackened box fitted with an observing tube T } in which 
an eyepiece E can be slid to focus upon the test plate 5. At B 
are the test characters, letters and figures bearing no relation to 
each other, and these are illuminated through the window W, of 
translucent material. This window can be varied in aperture by 
the shutter S, moved by a rack and pinion from the outside, where 
an arbitrary scale is provided. The window W is turned to the 
light to be examined and the shutter is then moved until the test 


characters are just visible, when the illumination can be read off 
upon the previously calibrated scale. 

Another instrument of similar character is the reading photom- 
eter shown in section in Fig. 33. This instrument is a blackened 
box mounted on a convenient handle and fitted with a wide eye 
hood jS and a short sight tube L on the rear side of the box; at C 
is a card bearing unrelated letters or characters. A variety of 
cards with various characters is provided to increase the range of 
the instrument and to prevent the observer from becoming ac- 
quainted with the characters. In use, the hood S is held to the 
eyes while the sight tube is pointed over the shoulder toward the 
light to be observed, and the observer then walks away until he 

Fig. 33. 

is unable longer to read the characters upon the card. The vanish- 
ing value of illumination for each card has to be determined as 
well as may be by a previous calibration, unless, as is more usual, 
the instrument is used for merely comparative purposes. 

As commonly worked, with cards and type so coarse as to bo 
dimly legible under an illumination of only a few thousandths of a 
foot-candle, this reading photometer is subject to so enormous a per- 
sonal equation as to be utterly unreliable, although if used by a 
single person thoroughly familiar with it and with considerably finer 
type than that customarily supplied, it is not wholly to be despised 
as an adjunct to the judgment. The type used in an instrument of 
this class should be at least as fine as ordinary newspaper type to 
avoid bringing the extinction value of the illumination so low as 


to compel the eye to work under conditions that make the result 
depend almost entirely on the state of adaptation of the eye and 
the imagination of the observer. 

The same criticisms hold to a somewhat less extent for the illu- 
minometer previously described and for all other instruments that 
depend on reading characters under greatly reduced illumination. 
Such instruments when carefully used within moderate range of 
intensities are sometimes very convenient and reasonably con- 
sistent, but the personal equation involved in their use is too large 
and too variable to render them generally trustworthy. 




At root, nearly all practical illuminants are composed of solid 
particles, usually of carbon, brought to vivid incandescence. We 
may, however, divide them into two broad classes, according as 
the incandescent particles are heated by their own combustion or 
by extraneous means. The first class, therefore, may be regarded 
as composed of luminous flames, such as candles, lamps, ordinary 
gas flames, and the like; while the second consists of illuminants 
in which a solid is rendered incandescent, it is true, but not by 
means of its own combustion. 

The second class thus consists of such illuminants as mantle 
gas burners, electric incandescent lamps, and the ordinary elec- 
tric arcs, which really give their light in virtue of the intense 
heating of the tips of the carbons by the arc, which in itself is 
relatively of feeble luminosity. 

Illumination based on incandescent gas, phosphorescence, and 
the like is in an early stage of development, and while it is in 
this direction that we must look for increased efficiency, incandes- 
cent illuminants are still the main reliance in artificial lighting. 
To the examination of flame illuminants, then, we must first 
address ourselves. 

They are interesting as being the earliest sources of artificial 
light, and, while usually of much less luminous efficiency than the 
second class referred to, still hold their own in point of conven- 
ience, portability, and ease of extreme subdivision. 

We have no means of knowing the earliest sources of artificial 
light as distinguished from heat. The torch of fat wood was a 
natural development from the fire on the hearth. But even in 
Homeric times there is clear evidence of fire in braziers for the 
purpose of lighting, and there is frequent mention of torches. 
The rope link saturated with pitch or bitumen was a natural 
growth from the pine-wood torch, and was later elaborated into 
the candle. 



It is clear that both lamps and candles date far back toward 
prehistoric times, the lamp being perhaps a little the earlier of 
the two. At the very dawn of ancient civilization man had 
acquired the idea of soaking up animal or vegetable fats into a 
porous wick and burning it to obtain light, and the use of soft 
fats probably preceded the use of those hard enough to form 
candles conveniently. 

The early lamps took the form of a small covered basin or 
jar with one or more apertures for the wick and a separate aper- 
ture for filling. They were made of metal or pottery, and by 
Roman times often had come to be highly ornamented. Fig. 34 

Fig. 34. — Early Roman Lamps. 

shows a group of early Roman lamps of common pottery, and 
gives a clear idea of what they were. They rarely held more 
than one or two gills, and must have given at best but a flicker- 
ing and smoky light. Fig. 35 shows a later Roman lamp of fine 
workmanship in bronze. 

In very early times almost any fatty substance that would burn 
was utilized for light, but in recent centuries the cruder fats have 
gone out of general use, and new materials have been added to 
the list. It would be a thankless task to tabulate the properties 
of all the solids and liquids which have been burned as illumi- 
nants, but those in practical use within the century just passed 
may for convenience be classified about as follows: 




Fats and Waxes. Fats and Waxes. 

Tallow (stearin). 
Sperm oil (whale oil). 
Lard oil. 


Olive oil. 
Colza oil. 
Vegetable waxes. 

The true fats are chemically glycerides, i.e., combinations of 
glycerin with the so-called fatty acids, mainly stearic, oleic, and 
palmetic. The waxes are combinations of allied acids with bases 

,*. »■ 

*^T*- ft 


Fig. 35. — Roman Bronze Lamp. 

somewhat akin to glycerin, but of far more complicated composi- 
tion. Technically, spermaceti is allied to the waxes, while some 
of the vegetable waxes belong chemically with the fats. 

All these substances, solid or liquid, animal or vegetable, are 
very rich in carbon. They are composed entirely of carbon, 
hydrogen, and oxygen, and as a class have about the following 
percentage composition by weight: carbon, 76 to 82 per cent; 
hydrogen, 11 to 13 per cent; oxygen, 5 to 10 per cent. 

They are all natural substances which merely require to go 


through a process of separation from foreign matter, and some- 
times bleaching, to be rendered fit for use. 

An exception may be made in favor of "stearin/' which is 
obtained by breaking up chemically the glycerides of animal 
fats and separating the fatty acids before mentioned from the 
glycerin. The oleic acid, in which liquid fats are rich, is also 
gotten rid of in the commercial preparation of stearin in order 
to raise the melting point of the product. 

In a separate class stand the artificial "burning fluids" used 
considerably toward the middle of the nineteenth century. As 
they are entirely out of use, they scarcely deserve particular 
classification. Their base was usually a mixture of wood alco- 
hol and turpentine in varying proportions. From its great 
volatility such a compound acted almost like a gas generator; 
the flame given off was quite steady and brilliant, with much 
less tendency to smoke than the natural oils, but the " burning 
fluids" as a class were outrageously dangerous to use, and for- 
tunately were driven out by the advent of petroleum and its 

Petroleum, which occurs in one form or another at many places 
on the earth's surface, has been known for many centuries, although 
not in large amounts until recently. Bitumen is often mentioned 
by Herodotus and other early writers, and in Pliny's time mineral 
oil from Agrigentum was even used in lamps. 

But the actual use of petroleum products as illuminants on a 
large scale dates from a little prior to 1860, when the American 
and Russian fields were developed with a common impulse. Crude 
petroleum is an evil-smelling liquid, varying in color from very 
pale yellow to almost black, and in specific gravity from 0.77 to 
1.00, ranging commonly from 0.80 to 0.90. 

Chemically it is composed essentially of carbon and hydrogen, 
its average percentage composition being about as follows : carbon, 
85 per cent; hydrogen, 15 per cent. It is composed in the main 
of a mixture of the so-called paraffin hydrocarbons, having the 
general formula C»H 2w +2, and the members of this series found 
in ordinary American petroleum vary from methane (CH 4 ) to 
pentadecane (Ci 6 H32), and beyond to solid hydrocarbons still more 
complicated. Petroleum from the Texas and neighboring fields 
and from the Russian fields is generally less rich in the paraffin 
series and contains members of other hydrocarbon series in consid- 



erable amounts, yielding interesting and valuable products, but 
less of high-grade illuminating oils. 

To fit petroleum for use as an illuminant, these component 
parts have to be sorted out, so that the oil for burning shall 
neither be so volatile as to have a dangerously low flashing point 
nor so stable as not to burn clearly and freely. 

This sorting is done by fractional distillation. The following 
tabic gives a general idea of the products arranged according to 
their densities: 

Petroleum ether. 

Petroleum spirit 





" Benzine naphtha, 


I Kerosene of vari- 
ous grades, 

Density. Use. 


q g« > Small, as solvents. 

0.65 Gas, explosion engines. 

0.68 Gas lamps, engines. 

0.71 Cleaning, engines. 

0.74 Varnish, etc. 

to (-Illumination. 

rui ( Lubricating oils }0 

Oils ] . ° t 

( of various grades, q 





S Candles, insulation, 
waterproofing, etc. 

" Petroleum ether" and "petroleum spirit " find little direct 
use in illumination, for they are so inflammable as to be highly 
dangerous, and form violently explosive mixtures with air at 
ordinary temperatures. 

Kerosene should be colorless, without a very penetrating odor, 
which indicates too great volatility, and should not give off in- 
flammable vapor below a temperature of 120° F., or, better still, 
below 140° F. to 150° F. Oils of the latter grades are pretty 
safe to use, and are always to be preferred to those more volatile. 
The yield of kerosene from crude oil varies from place to place, 
but with good American oil runs as high as 50 to 75 per cent. 

Paraffin is sometimes used unmixed for making candles, but 
is preferably mixed with other substances, like stearin, to give 
it a higher melting point. 

Having thus casually looked over the materials burned in candles 
and lamps, the results may properly be considered. 

Candles. — These are made usually of stearin, paraffin, wax, or 
mixtures of the first two substances. They are molded hot in 


automatic machines, and, as usually supplied in this country, 
are made in weights of 4, 6, and 12 to the pound. Spermaceti 
candles are also made, but are little used except for a standard 
of light. The old English standard candle is of spermaceti, weigh- 
ing one-sixth of a pound and burning at the rate of 120 grains per 

Commercial candles give approximately 1 candle power, some- 
times rather more, and burn generally from 110 to 130 grains per 
hour. As candles average from 15 to 18 cents per pound, the cost 
of 1 candle hour from this source amounts to about 0.25 cent to 
0.30 cent. This is obviously relatively very expensive, although it 
must not be forgotten that candles subdivide the light so effectively 
that for many purposes 16 lighted candles are very much more 
effective in producing illumination than a gas flame or incandescent 
lamp of 16 candle power. 

The present function of candles in illumination is confined to 
their use as portable lights, for which, on the score of safety, they 
are far preferable to kerosene lamps, and to cases in which, for 
artistic purposes, thorough subdivision of light is desirable. Where 
only a small amount of general light is needed, candles give a 
most pleasing effect, and are, moreover, cleanly and odorless. 

In efficiency candles leave much to be desired. For, taking the 
ordinary stearin candle as a type, it requires in dynamical units 
the equivalent of about 90 watts per candle power, consumes per 
hour the oxygen contained in 4.5 cubic feet of air, and gives off 
about 0.6 cubic foot of carbonic acid gas. In these respects the 
candle is inferior to the ordinary lamp, and still more inferior to 
gas or electric lights. Nevertheless, it is oftentimes a most con- 
venient illuminant. 

Oil Lamps. — Oils other than kerosene are used in this country 
only to a very slight extent, the latter having driven out its com- 
petitors. Sperm oil and, abroad, colza oil (obtained from rapeseed) 
are valued as safe and reliable illuminants for lighthouses, and in 
some parts of the Continent olive oil is used in lamps, as it has been 
from time immemorial. 

Here, kerosene is still the general illuminant outside of the cities 
and larger towns. It has the merits of being cheap (on the 
average 12 cents to 15 cents per gallon in recent years), safe, if 
of the best quality, and of giving, when properly burned, a very 
steady and brilliant light. 


All oils require a liberal supply of air for their combustion, par- 
ticularly the heavier oils, and many ingenious forms of lamp have 
been devised to meet the requirements. On the whole, the most 
successful are on the argand principle, using a circular wick with 
air supply both within and without, although some of the double 
flat-wick burners are admirable in their results. A typical burner, 
the familiar " Rochester," is shown in Fig. 36, which sufficiently 
shows the principle involved. In kerosene lamps the capillary 
action of the wick affords an ample supply of oil, but with some 
other oils it has proved advantageous to provide a forced supply. 

Fig. 36. — "Rochester" Kerosene Burner. 

The so-called "student lamp," with its oil reservoir, is the survival 
of an early form of argand burner designed to burn whale oil, and 
gives a particularly fine and steady light. In other instances clock- 
work is employed to pump the oil, and sometimes a forced-air 
supply is UBed. 

Kerosene lamps usually arc designed to give from 10 to 20 
candle power, and occasionally more, special lamps giving even 
up to 50 or 60 candle power. The consumption of oil is generally 
from 50 to 60 grains per hour per candle power. As kerosene 
weighs about 6.6 pounds per gallon, the fight obtained is in the 
neighborhood of 800 candle hours per gallon. 

This brings the cost of the candle hour down to about 0.018 


cent for material consumed, taking the oil at 15 cents per gallon. 

No illuminants save arc lights, metallic filament incandescents, 
and mantle burners with cheap 
gas can compare with it in point 
of economy. 

A very interesting and valuable 
application of oil lighting is found 
in the so-called " Lucigen " torch 
and several kindred devices. The 
oil, generally one of the heavier 
petroleum products, is carried 
under air pressure in a good-sized 
portable reservoir, and the oil is 
led, with the compressed air 
strongly heated by its passage 
through the apparatus, to an 
atomizing nozzle, from which it 
is thrown out in a very fine spray, 
and is instantly vaporized and 
burned under highly efficient con- 

These " Lucigen" torches give 
nearly 2000 candle power on a 
consumption of about two gallons 
of oil per hour, burning with a 
tremendous flaring flame three 
feet or more in length and six or 
eight inches in diameter. They 
are very useful for lighting ex- 
cavations and other rough works 
for night labor, being powerful, 
portable, and cheap to operate. 
Fig. 37 gives an excellent idea of 
this apparatus in a common form. 

Pig. 37. — "Lucigen" Torch. Sucn a li 8 ht ta 8uited onIv to 

outdoor work, but it forms an 
interesting transitional step toward the air-gas illuminants which 
have come into considerable use for lighting where service mains 
for gas or electricity are not available, or where the conditions 
confer special economy. 



Air Gas. — It has been known for seventy years or more that 
the vapor of volatile hydrocarbons could be used to enrich poor 
coal gas, and that even air charged with a large amount of such 
vapor was a pretty good illuminant. 

Of late years this has resulted in the considerable use of "car- 
buretors," which saturate air with hydrocarbon vapor, making 
a mixture too rich to be readily explosive and possessing good 
illuminating properties when burned as gas in the ordinary way. 

Fig. 38. — Gasoline Gas Machine. 

The usual basis of operations is commercial gasoline, which 
consists of a mixture of the more volatile paraffin hydrocarbons, 
chiefly pentane, hexane, and isohexane. 

The process of gas making is very simple, consisting merely 
of charging air with the gasoline vapor. Fig. 38 shows in section 
a typical air-gas machine. It consists of a large metal tank hold- 
ing a supply of gasoline, a carbureting chamber of flat trays over 
which a gasoline supply trickles, a fan to keep up the air supply, 
and a little gas reservoir in which the pressure is regulated and 


from which the gas is piped. The fan is usually driven by heavy 
weights, wound up at suitable intervals. 

The whole gas machine is usually put in an underground 
chamber, both for security from fire and to aid in maintaining 
a steady temperature. About six gallons of gasoline are required 
per 1000 cubic feet of air, and the result is a gas of very fair 
illuminating power, rather better than ordinary city gas. 

The cost of this air gas is very moderate, but on account of the 
cost of plant and some extra labor, it is materially greater than 
the cost of direct lighting by kerosene lamps. It is a means of 
lighting very useful for country houses and other places far from 
gas or electric supply companies. The principal difficulty is the 
variation of the richness of the mixture with the temperature, owing 
to change in the volatility of the gasoline, a fault which is very 
difficult to overcome. At low temperatures there is a tendency to 
carburet insufficiently and to condense liquid in the cold pipes. 
The gas obtained from these machines is burned in the ordinary 
way, although burners especially adapted for it are extensively 
employed. In recent years such gas has been considerably used 
with mantle burners, obtaining thus a very economical result. The 
air gas just described is too rich in gasoline vapor to be an explosive 
mixture, the limits of danger being between 2 and 5 per cent of 
gasoline vapor, and the mixture described being well above the 
upper limit. Abroad air gas so lean as to be below the lower limit 
of danger has come into use, carrying say 1.5 per cent gasoline 
vapor instead of 10 to 15 per cent. Such gas is fit only for use 
in mantle burners, but is cheap and safe. 

Coal Gas. — In commercial use for nearly a century, coal gas 
was, until about twenty-five years ago, the chief practical illumi- 
nant. Little need here be said of its manufacture, which is a 
department of technology quite by itself, other than that the gas 
i3 obtained from the destructive distillation of rich coals inclosed 
in retorts, from which it is drawn through purifying apparatus and 
received in the great gasometers familiar on the outskirts of every 

The yield of gas is about 10,000 cubic feet per ton of coal of 
good quality. The resulting gas consists mainly of hydrogen and 
of methane (CH 4 ), with small amounts of other gases, the composi- 
tion varying very widely in details while preserving the same 
general characteristics. A typical analysis of standard coal gas 


giving 16 to 17 candle power for a burner consuming 5 cubic feet 
per hour would be about as follows: 

Hydrogen 63 .0 

Paraffin hydrocarbons 33 .0 

Other hydrocarbons 3.5 

Carbon monoxide s 5.5 

Carbon dioxide 0.6 

Nitrogen f 4.2 

Oxygen 0.2 


Ammonia compounds, carbon dioxide, and sulphur compounds 
are the principal impurities which have to be removed. Traces of 
these and of moisture are regularly found in commercial gas. Sul- 
phur dioxide (SO2) is the most persistent impurity and perhaps the 
most objectionable. 

In point of fact, at the present time but a small proportion of 
the illuminating gas used in this country is unmixed coal gas, such 
as might show the analysis just given. Most of it is water gas, 
or a mixture of coal gas and water gas. Water gas is produced 
by the simple process of passing steam through a mass of incandes- 
cent coal or coke, and thus breaking up the steam into hydrogen 
and oxygen, which latter unites with the carbon of the coal, form- 
ing carbon monoxide. 

At moderate temperatures considerable carbon dioxide would be 
formed, but, as this is worse than useless for burning purposes, the 
heat is always carried high enough to insure the formation of the 
monoxide. The hypothetical chemical equation is: 

H 2 + C = CO + H 2 . 

The reaction is never clean in so complete a sense as this, some 
CO2 always being formed. This water gas as thus formed is useless 
as an illuminant, and requires to be enriched by admixture of light- 
producing hydrocarbons — carbureted, in other words. This is 
done by treating it to a spray of petroleum in some form, and at 
once passing the mixture through a superheater, which breaks down 
the heavier hydrocarbons and renders the mixture stable. 

There are many modifications of this system worked on the 
same general lines. The enriching is carried to the extent neces- 
sary to meet the legal requirements, usually producing gas of 15 
to 20 candle power for a 5-foot jet. A typical analysis of the 
water gas after enriching would show about the following by 


Hydrogen 34.0 

Methane 15.0 

Enriching hydrocarbons 12 . 5 

Carbon monoxide 33 .0 

Oxygen, nitrogen, C02, etc 5.5 


The latter part of the enriching process, i.e., superheating and 
breaking up the heavy hydrocarbons while in the form of vapor, 
is substantially that used in making Pintsch and allied varieties 
of oil gas, so that commercial water gas may be regarded as a 
mixture of water gas and oil gas. The "cracking " of heavy oils 
by heat has proved a convenient means of increasing the avail- 
able amount of lighter hydrocarbons from petroleum. The lightest 
gaseous products are sometimes separated and compressed to lique- 
faction in steel cylinders, thus furnishing an easily transportable 
and convenient source of gas for light and heat. Such is the 
so-called " Blau-gas " and some analogous products in commercial 
use to a limited extent. 

Water gas, when properly enriched, is fully the equivalent of 
coal gas for illuminating purposes. The main difference between 
them is the very large proportion of carbon monoxide in the water 
gas, which adds greatly to the danger of leaks. 

For carbon monoxide is an active poison, not killing merely 
by asphyxia, but by a well-defined toxic action peculiar to itself. 
Hence persons overcome by water gas very frequently die under 
circumstances which, if coal gas were concerned, would result 
only in temporary insensibility. As the enriched water gas is 
cheaper than coal gas, however, the gas companies, maintaining, 
with some justice, that gas is not furnished for breathing pur- 
poses, supply it unhesitatingly — sometimes openly, sometimes 
without advertising the fact. 

Very commonly so-called coal gases contain enriched water gas 
to bring up their illuminating power. In these cases the carbon 
monoxide is in much less proportion, perhaps only 10 to 15 per 

It is often stated that water gas is doubly dangerous from 
its lack of odor. The unenriched gas is practically odorless, but 
when enriched the odor, while less penetrating than that of coal 
gas, is sufficiently distinctive to make a leak easily perceptible. 

Gas burners for ordinary illuminating gas are of three general 
types: flat flame, argand, and regenerative. The first named is 


the most common and least efficient form. It consists of two 
general varieties, known respectively as the "fishtail" and "bat's- 
wing." The former has a concave tip, usually of steatite or sim- 
ilar material, containing two minute round apertures, so inclined 
that the two little jets meet and flatten out crosswise into a wide 
flame. This form is now relatively little used save in dealing 
with some special kinds of gas. 

The bat's-wing burner, with a dome-shaped tip, having a narrow 
slit for the gas jet, is the usual form employed with ordinary 
gas. Flat-flame burners work badly in point 
of efficiency unless of fairly large size. On 
ordinary gas of 14- to 17-c.p. nominal value 
on a 5-foot burner, burners taking less than 
about 4 cubic feet per hour are decidedly in- 
efficient. A 4-foot burner will give about 
2.5 candles per foot, while a 5-foot burner 
will give 2.75 to 3 candle power per foot. 

The argand burners give considerably 
better results, their flames being inclosed 
and protected from draughts by a chimney; 
and the air supply being good the tem- 
perature of the flame is high and the light 
is whiter than in the flat-flame burners. 
The principle is familiar, the wick of the 
argand oil lamp being replaced in the gas 
burner by a hollow ring of steatite connected 
with the supply, and perforated with tiny 
jet holes around the upper edge. Pig. 39 
shows in section an argand burner (Sugg's) 
of a standard make used in testing London 
gas. This burner uses 5 cubic feet per hour, 
and the annular chamber has 24 holes, each 
0.045 inch in diameter. The efficiency is a 
little better than that of the flat-flame Fig. 39. — Section of 
burners, running, on good gas, from 3 to Ar K and Gas Buroor. 
3.5 candle power per foot. The London legal standard gas is of 
16 candle power in this 5-foot burner. 

On rich gas the flat-flame burners, particularly the fishtail, 
work better than the argand, the fishtail being better on very 
rich gas than is the bat's-wing form. With ordinary qualities 


of gas, however, the argand burner is vastly more satisfactory 
than the flat flames. 

For very powerful lights the so-called regenerative burners are 
generally preferred. These are based on the general principle of 
heating both the gas and the air furnished for 
its combustion prior to their reaching the 
flame. The burner proper is something like 
an inverted argand, so arranged as to furnish 
a circular sheet of flame turned downward, and 
with, of course, a central cusp. Directly above 
the burner, and strongly heated by the flame, 
are the air and gas passages. 

Fig. 40 shows in section the Wenham burner 
of this class. The arrows show the course of 
the air and the gas, the latter being burned 
just below the iron regenerative chaml>er and 
the products of combustion passing upward 
through the upper shell of the lamp, and 
preferably to a ventilating flue. The globe 
below prevents the access of cold air, and an 
annular porcelain reflector surrounding the exit 
flue turns downward some useful light. 

The Siemens regenerative burner, arranged 
upon a 'similar plan and shown in Fig. 41, 
gives much the same effect. The regenerative 
Fig. 40. — Wenham burners of this class give a very brilliant yellow- 
Regenerative Burner. wmte | ignt ^j, a genera Hy hemispherical dis- 
tribution downward. They work best and most economically in 
the larger sizes, — . 100 to 200 candle power, — and must be placed 
near the ceiling to take the best advantage of their usual dis- 
tribution of light. 

With gas of about 16-candlo-power standard these regenerative 
burners consume only about 1 cubic foot per hour for 7 to 10 
candle power. They are thus more than twice as economical as 
the best argand burners. Their chief disadvantage lies in the 
fact that to get this economy very powerful burners must be 
used, of a size not always conveniently applicable. 

From such a powerful center of light a large amount of heat 
is thrown off, obviously less per candle power of light than 
in other gas burners, but, in the aggregate, large. Regenerative 


burners have done good service in the illumination of large 
spaces, although at the present time the greater economy of the 
mantle burner has pushed the regenerative class into the back- 
ground. Their light, neverthe- 
less, is of a more pleasing color 
than that given by the mantle 

The most recent and in some 
respects most important ad- 
dition to the list of Same illu- 
minants is acetylene. This gas 
is a hydrocarbon having the ' 
formula CjHj, which has been 
well known to chemists for 
many years, but which until 
recently has not been prepar- 
able by any convenient com- 
mercial process. It is a rather 
heavy gas, of evil odor, gener- 
ally somewhat reminiscent of 
garlic, and, being extremely 
rich in carbon uncombined with 
oxygen (nearly 93 per cent by 
weight), it bums very bril- 
liantly when properly supplied 
with air. Its flame is intensely 
bright, nearly white in color, 
' and for the light given it viti- 

ates the air in comparatively Rg.41. -Semen. Regenerativi, 
small degree. GaH Burner. 

Acetylene is made in prac- 
tice from calcic carbide, CaCj, a chemical product prepared by 
subjecting a mixture of powdered lime and carbon (coke) to the 
heat of the electric furnace. By this means it can be prepared 
readily in quantity at moderate cost. The acetylene is made 
from the calcic carbide by treating it with water, lime and acety- 
lene being the results of the reaction, which, in chemical terms, is 
as follows: 

CaCs + 2 H,0 = Ca(OH), + C,H*. 



Commercial calcic carbide is far from being chemically pure, so 
that the acetylene prepared from it contains various impurities, 
and is neither in quantity nor quality just what the equation would 
indicate. The carbide is extremely hygroscopic, and hence not 
very easy to transport or keep, and the upshot of this property 
and the inherent impurities is that the practical yield of acetylene 
is only about 4.5 to 5.0 cubic feet per pound of carbide, 4.75 
cubic feet being an extremely good average unless the work is 
on a very large scale, though 4.5 cubic feet is the more usual 
yield. In theory the yield should be nearly 5,5 cubic feet per 

The gaseous impurities are quite varied and by no means uni- 
form in amount or nature, but the most objectionable ones may 
be removed by passing the gas in fine bubbles through water. 
If the gas is being prepared on a large scale it can readily be 

Acetylene has the disadvantage of being somewhat unstable. 
It forms direct compounds with certain metals, notably copper, 
these compounds being known as acetylides, and being themselves 
so unstable as to \>e easily explosive. Acetylene should be there- 
fore kept out of contact with copper in storage, and even in 

The gas itself is easily dissociated with evolution of heat into 
carbon and hydrogen, and hence may be inherently explosive 
under certain conditions, fortunately not common. 

At atmospheric pressure, or at such small increased pressures 
as are employed in the commercial distribution of gas, acetylene, 
unmixed with air, cannot be exploded by any means ordinarily 
at hand. 

Above a pressure of about two atmospheres acetylene is readily 
explosive from high heat and from a spark or flame, and grows 
steadily in explosive violence as the initial pressure rises, until 
when liquefied it detonates with tremendous power if ignited. 
At ordinary temperatures it can be liquefied at a pressure of 
about 80 atmospheres, and it has been proposed to transport 
and store it in liquid form. But, although even when liquefied 
it will not explode from mechanical shock alone, it is in this 
condition an explosive of the same order of violence as guncotton 
or nitroglycerin, and should be treated as such. 

Mixtures of acetylene and air explode violently, just as do 


mixtures of illuminating gas and air. The former begin to explode 
rather than merely burn, when the mixture contains about one 
volume of acetylene to -three of air, detonate very violently with 
about nine volumes of air, and cease to explode with about twenty 
volumes of air. 

Ordinary coal gas begins to explode when mixed with three 
volumes of air, reaches a maximum of violence with about five 
to six volumes, and ceases to explode with eleven volumes. Of 
the two gases, the acetylene is rather the more violently explosive 
when mixed with air, and it becomes explosive while the mixture 
is much leaner. The difference is not of great practical moment, 
however, except as acetylene generators, being easily operated, 
are likely to get into unskillful hands. This fact has already 
resulted in many disastrous explosions. 

As regards its poisonous properties, acetylene seems to be some- 
what less dangerous than coal gas and very much less dangerous 
than water gas. Properly speaking, acetylene is but very feebly 
poisonous when pure, and has such an outrageous smell when 
slightly impure that the slightest leak attracts attention. Some 
early experiments showed highly toxic properties, but these have 
not been fully confirmed, and may have been due to impurities 
in the gas — possibly to phosphine, which is a violent poison. 

The calcic carbide from which the acetylene is prepared is so 
hygroscopic and gives off the gas so freely that it has to be stored 
with great care on account of possible danger from fire. Fire under- 
writers are generally united in forbidding entirely the use or storage 
of liquid or compressed acetylene, or the storage of any but trivial 
amounts of calcic carbide (a few pounds) except in detached fire- 
proof buildings. 

Acetylene is, when properly burned, a magnificent illuminant. 
It will not work in ordinary burners, for unless very liberally sup- 
plied with air it is so rich in carbon as to burn with a smoky flame 
and a deposit of soot. It must actually be mixed with air at the 
burner in order to be properly consumed. When so utilized its 
illuminating power is very great. The various experiments are not 
closely concordant, but they unite in indicating an illuminating 
power of 35 to 50 candle hours per cubic foot, according to the 
capacity of the burner, the larger burners, as usual, working the 
more economically. 

This means that the acetylene has nearly fifteen times the 



illuminating power of a good quality of ordinary illuminating gas 
when burned in ordinary burners. It will, consequently, give 
about six to eight times more light per cubic foot than gas in a 
regenerative burner, and, it may be mentioned, about three to four 
times more light than gas in a mantle (Welsbach) burner. 

Fig. 42 shows a common standard form of acetylene burner, 
intended to consume about 0.5 cubic foot per hour. It is a duplex 

form akin in its production of flame to a 
common fishtail. Each of the two burners 
is formed with a lava tip having a slight 
constriction close to its point. In this is 
. the central round aperture for the gas, and 
just ahead of it are four lateral apertures 
for the air supply. The acetylene and air 
mix just in front of the constriction and 
the two burners unite their jets to form 
a small, flat flame. It is in effect a pair 
of tiny Bunsen burners inclined to produce 
a fishtail jet. 

Larger acetylene burners are worked on 
a similar principle, all having the air- 
supply passages characteristic of the Bun- 
sen burner. Too great air supply for the acetylene gives the 
ordinary colorless Bunsen flame, but on reducing the amount the 
acetylene burns with a singularly white, brilliant, and steady flame. 
Of acetylene generators designed automatically to supply gas 
at constant pressure from the calcic carbide the name is legion. 
A vast majority of those in use at present are of rather small 
, capacity, being designed for a few lights locally or as portable 
apparatus for lamps used for projection. 

A very useful type of the small generator is shown in Fig. 43, 
a form devised by d'Arsonval. It consists of a small gasometer 
with suitable connections for taking off the gas and drawing off 
the water. The bell of the gasometer is furnished at the top with 
a large aperture closed by a water seal. Through this is intro- 
duced a deep iron-wire basket containing the charge of carbide. 

The acetylene is generated very steadily after the apparatus 
gets to working and the pressure is quite uniform. The water in 
the gasometer of the d'Arsonval machine is covered by a layer of 
oil, which serves an. important purpose. When one ceases using 

Fig. 42. — Acetylene 


the gas the bell rises, and as the carbide basket rises out of the 
water the oil coats it and displaces the water, checking further 
evolution of gas. The oil also checks 
evaporation, so that there is no slow evo- 
lution of gas from the absorption of 
aqueous vapor. 

Acetylene generators on a larger scale 
arc operated on much the same principle, 
although the generating and regulating 
parts of the apparatus are commonly sepa- 
rated instead of being united as in Fig. 
43, and there is often means for washing 
the gas. The principle adopted in the 
larger generators is uniformly to feed the 
calcic carbide in small quantities in a 
large excess of water, thereby avoiding 
the overheating which would follow were 
water dropped on an excess of carbide. 

Fig. 44 shows in section a typical 
acetylene apparatus for medium-sized 
plants such as used for house lighting. 
It consists of a generator tank, a carbide 
holder from which the granulated car- 
bide is automatically fed in small quan- 
tities, a regulating gasometer, and the R 8- *3. — Small Acetylene 
necessary piping, valves, and water seal encw r ' 

to facilitate and insure the safe and convenient handling of the 
output. Such apparatus runs with very little attention and is 
largely used in country houses and hotels out of reach of gas and 
electric supply, and even as a matter of economy where the gas 
supply is indifferent in quality or too high in price. 

Isolated plants of capacity as high as 5000 burners have been 
installed for service in large hotels. Occasionally a fairly large 
plant is found in village lighting, but for the most part the equip- 
ments as generally found are for a few score or hundred burners. 
These acetylene generators as now furnished are very nearly auto- 
matic in their action, but at times require a little intelligent care to 
keep them in first-class working order, as they are not yet entirely 
foolproof, so that while the labor and upkeep cost is small it is not 


Acetylene has a calorific value of about 1440 B.t.u. and has been 
successfully applied to the production of enormously high tempera- 
tures by burning it with oxygen as a substitute for the older 
oxyhydrogen flame. Used in this way, it gives a blowpipe flame 
capable of most valuable service in welding, cutting steel plates 
and beams, and other sensational feats requiring extreme tempera- 
ture. Even the ordinary flame temperature of acetylene is in the 
neighborhood of 2200° C. 

Fig. 44. — Type of Acetylene Generator Used for Mediiun-siied Plants 

and House Lighting. 

Acetylene is particularly well adapted for supplying portable 
lights in virtue of the great ease with which it can be stored in 
solution. It is not a good thing to store under simple pressure, 
but it can be stored in solution very simply and safely. Acetone 
absorl>s acetylene greedily, aa water takes up ammonia, so that 
at ordinary temperature and pressure 25 volumes of gas are taken 
up by one volume of liquid, and under pressure proportionately 

To avoid the inconvenience of dealing with a liquid and to 
increase security against explosion, the pressure cylinders used 
for the acetone solution are filled with very porous asbestos disks 


which still leave 80 per cent of the real capacity for the acetylene- 
acetone solution, which is carried up to a pressure of about 
ten atmospheres. A cylinder so charged contains an amount of 
acetylene equivalent at atmospheric pressure to 100 times its 
volume. Hence a cylinder of no more than half a cubic foot 
capacity can supply a 40-candle-power burner for something like 
100 hours of continuous service. Acetylene thus stored has come 
into very large use for lighting motor cars, portable lamps, buoys, 
and other purposes where a convenient and powerful portable 
source of light is needed. 

At the present time, in fact, acetylene is practically the only 
flame illuminant which must, be seriously taken into consideration 
by the illuminating engineer. For furnishing portable lights of 
considerable power it is easily the most convenient source avail- 
able, and in an emergency a good automobile light burning 
acetylene from the usual storage tanks is capable of doing im- 
mensely efficient service in lighting up night work of various kinds; 
and it is available in any city on extremely short notice. 

As to the value of acetylene, it is evidently worth about fifteen 
times as much per cubic foot as gas burned in ordinary burners, 
or three to four times as much as gas, assuming the latter to be 
burned in Welsbach burners. Now, one ton of calcic carbide 
of high quality, efficiently used, will produce between 9000 and 
10,000 cubic feet of acetylene, equal in illuminating value to 
150,000 cubic feet of gas in the one case or to 30,000 to 40,000 
cubic feet in the other. 

The cost of the calcic carbide is a very uncertain quantity at 
present. The best authorities bring the manufacturing cost, on 
a large scale and under very favorable circumstances, somewhere 
between $30 and $40 per ton. It is doubtful if any finds its way 
into the hands of bona-fide users at less than about $60 per ton, 
and the current price in small lots is much higher, and naturally 
so, by reason of troublesome storage and the cost of transporta- 
tion. Adding the necessary allowance for the cost of producing 
the gas from the carbide, it is at once evident that the cost of 
lighting by acetylene falls materially below that of lighting by 
common gas in ordinary burners at the common price of $1 to 
$1.50 per 1000 feet. 

It is equally evident that it considerably exceeds the cost of 
gas lighting by Welsbach burners. Acetylene is to some extent 


used in mantle burners, but it is a question whether the moderate 
increase in efficiency overbalances the practical difficulties which 
have been found. Its cost of production and distribution does 
not yet render it commercially attractive under ordinary condi- 
tions of supply from gas and electric central distributing plants. 

Nevertheless, acetylene is, for use in isolated places, one of the 
very best and most practical illuminants; for it is fairly cheap, 
easily made, and gives a light not surpassed in quality by any 
common artificial illuminant save the " intensified " electric arc. 
It is peculiarly well adapted for temporary and portable use, 
giving as it does a very brilliant and steady light, well suited for 
use with reflectors and projecting apparatus, admirable in color, 
and very easy of operation. 




The general class of illuminants operative by the incandescence 
of a fixed solid body would include in principle both arc and 
incandescent electric lamps, as well as those in which the radiant 
substance is heated by ordinary means. In this particular place, 
however, it seems appropriate to discuss the latter forms only, 
leaving the electric lights for a separate chapter. 

Incandescent radiants brought to the necessary high tempera- 
ture by a nonluminous flame have their origin in the so-called 
"Drummond" or "lime" light, which has been used for many 
years as the chief illuminant in projection, scenic illumination 
on the stage, and such like purposes, and which has only recently 
been extensively replaced by the electric arc. The limelight con- 
sists of a short pencil of lime against which is directed the color- 
less and intensely hot flame from a blast lamp fed with pure 
oxygen and hydrogen, or more commonly with oxygen and illu- 
minating gas. 

The general arrangement of the oxyhydrogen burner is shown 
in Fig. 45. Here A and B are the supply pipes for the oxygen 
and hydrogen, fitted with stopcocks. These unite in a common 
jet in the burner C, which is usually inclined so as to bring the 
burner where it will not cast a shadow. Sometimes the two gases 
are mixed in the burner tube C, and sometimes the hydrogen is 
delivered through an annular orifice about a central tube which 
supplies the oxygen. The pencil of lime is carried on a holder Z>, 
and the whole burner is often carried on an adjustable stand E f 
so that it can be raised, lowered, or turned, as occasion demands. 
The mixed gases unite in a colorless, slender flame of enormously 
high temperature, and when this impinges on the lime the latter 
rises in a small circular spot to the most brilliant incandescence, 
giving an intense white light of, generally, 200 to 400 candle power. 

The light, however, falls off in brilliancy quite rapidly, par- 
ticularly when the initial incandescence is very intense, losing 



something like two-thirds of its candle power in an hour, so 
that it is the custom for the operator to turn the pencil from 
time to time so as to expose new portions to the oxyhydrogen 

At the highest temperatures the calcium oxide is somewhat vol- 
atile and the surface seems to change and lose its radiative power. 
Sometimes pencils of zirconium oxide are used instead of lime, 

Pig. 45. — Oxyhydrogen Burner. 

and this substance has proved more permanently brilliant and 
does not seem to volatilize. When properly manipulated, the 
calcium light is beautifully steady and brilliant, and being very 
portable, is well adapted for temporary use. 

From time to time attempts were made to produce a generally 
useful incandescent lamp in which the oxyhydrogen jet should 
be replaced by a Bunsen burner requiring only illuminating gas 
and air. 


Platinum gauze and other substances were tried as the incan- 
descent materials, but the experiments came to nothing practi- 
cally until the mantle burner of Auer von Welsbach appeared. 
This is generally known in this country as the Welsbach light, 
but on the Continent as the Auer light. In this burner the 
material brought to incandescence is a mantle, formed like a little 
conical bag, of thin fabric thoroughly impregnated with the proper 
chemicals and then ignited, leaving a coarse gauze formed of the 
active material. 

The composition of this material has been kept more or less 
secret until recently, and has been varied from time to time as 
the burner has gradually been evolved into its present state, but 
has always consisted of the oxides of the so-called "metals of the 
rare earths" and is actually composed of thoria with a minute 
percentage of ceria. 

These rare earths — zirconia, thoria, glucina, yttria, ceria, and a 
half-dozen others still less well known — form a very curious group 
of chemical substances. They are whitish or yellowish very refrac- 
tory oxides occurring as components of certain rare minerals, and 
most of them rise to magnificent incandescence when highly 
heated. The hue of this incandescence differs slightly for the 
different earths, and they are only slightly volatile at any but very 
high temperatures. One — erbia — has the property of giving a 
spectrum of bright bands when highly heated instead of the con- 
tinuous spectrum usual to incandescent solids, — a property which 
is shared in less degree by a few of its curious associates. 

The present composition of the Welsbach mantle, which is 
ordinarily about 99 per cent thoria and 1 per cent ceria, is the 
result of long-continued work over this group of substances by 
Dr. Auer von Welsbach, who found in purifying thoria for use in 
mantles that when chemically pure it gave on incandescence very 
little light. Tracing backward from this point to discover the 
properties of the impurities which he had gradually eliminated, 
he found that ceria possessed an extraordinary power of exciting 
the thoria to most brilliant incandescence. Just how this remark- 
able effect is produced is a mystery not yet fully fathomed. 
Dr. Auer's own idea is that the ceria, which is easily oxidized and 
reduced, in virtue of this property acts as a species of molecular 
excitant, perhaps going into combination with the thoria at one 
stage of oxidation and separating at another stage. Whether this 



explanation be the final one or not, it is at least a fact that in all 
cases, and they are quite numerous, in which a small percentage of 
one body confers the power of intensive radiation on a large per- 
centage of another body, the former is a substance readily assum- 
ing several stages of oxidation, and the latter has in general only 
one stage of oxidation and is highly refractory. At all events, in 
the Welsbach mantle and in other similar cases one always finds a 
fireproof oxide combined with an analogous oxide which is a facile 
oxygen carrier. It has been suggested that the two unite to form 
a sort of solid solution, and some plausibility is lent to this hy- 
pothesis by the curious sensitiveness of the thoria-ceria combination 
to small changes in the amount of ceria. As opposed to this cataly- 

Fig. 46. 

5 G 





Effect on the Candle Power of a Welsbach Mantle of Varying 

the Amount of Ceria. 

tic theory, some recent investigators hold that the action involved 
is purely physical, the ceria being an extraordinarily good and 
somewhat selective radiator, which by itself cannot be readily 
pushed to sufficiently high incandescence for practical efficiency, 
but which when mixed with the thoria of the mantle, itself a poor 
radiator but highly refractory, is heated to extreme incandescence. 
According to this view, it is practically the ceria only which gives 
the brilliancy to the mantle. There are objections to both theories, 
and the matter must be considered still unsettled. 

Fig. 46, due to Professor Whitaker, shows the effect on the 
candle power of a Welsbach mantle of varying the amount of ceria. 
The highest efficiency is reached with a percentage of ceria very 
close to one per cent, which is the amount commonly found in com- 




mcrcial mantles. With less than one-half per cent the efficiency 
falls off very rapidly indeed, and noticeably, although more slowly, 
with amounts of more than one per cent. To the illuminating 
engineer the theory of the Welsbach mantle is of less importance 
than its practice, which has completely revolutionized the gas in- 
dustry within the last decade or two. The standard Welsbach 
burner as it has been known with- 
in this period is shown in Fig. 47, 
in which the several parts arc 
plainly labeled. It consists es- 
sentially of a Bunsen burner with 
provisions for regulating the air 

and gas supplies, and the mantle tupport 

which surrounds the B unsen flame . 
There is a gauze tip on the Bun- 
sen burner to prevent the 0ame 
striking back, and suitable sup- 
ports for the shade, chimney, and 
mantle. ■ The mantle support is 
permanently adjusted to a cap 
with a wire-gauze top which goes 
upon the burner tube with a bayo- 
net joint so that the mantle is 
brought exactly to the right posi- 
tion. This standard burner uses 
txrtween 3.5 and 4 cubic feet of 
gas per hour, and gives with a 
first-class mantle about 60 mean 
spherical candle power when the 
mantle is new and of good quality. 

fore, in the neighborhood of 15 candle power per foot of gas. It 
is, therefore, four or five times as efficient as an open gas burner 
and two or three times as efficient as the best of the regenerative 
gas burners, that were, up to the coming of the Welsbach, the 
most effective gas illuminants. For a long time the Welsbach 
burner was only available in this unnecessarily powerful form. It 
consumed, to be sure, less gas than an ordinary fishtail burner, 
but it gave a great deal more light than was really wanted for most 
purposes, particularly in domestic lighting, so that the consumer, 
although getting very cheap light, would have been better satisfied 

Fig. 47. 
The initial efficiency is, there- 


with a smaller consumption of gas and a quantity of light bettor 
suited to his needs. 

In response to this demand, there has been made available 
within the past three years what is known to the trade as the 
Junior Welsbach, which is a very much smaller and simpler type 
of the same structure. It is well shown in Fig. 48. 
It consists of a Bunsen tube like the larger burner, with 
a very simple regulator carrying a tube supporting a 
short perforated mica chimney bearing the mantle 
already attached in position. The mantle itself is 
barely 2 inches long and the whole affair about 4 
inches, so that it screws on in place of an ordinary 
burner tip and falls inside the shades previously used 
for open flames. Chimney, support, and mantle are 
removed and thrown away when used up, and the total 
cost of replacement is practically the same as that of 
the mantle alone in the larger type of burner. These 
little burners are fully as efficient as their larger prede- 
^' *■ cessofs, and with new mantles give between 30 and 
35 mean spherical candle power on a consumption of 1.75 to 2 
feet of gas per hour, assuming gas of ordinary quality and pres- 
sure. In other words, their initial efficiency is in the neighborhood 
of 15 candle power per cubic foot of gas as in the case of the 
larger burners. This high figure in so small a burner is due prob- 
ably to improvements in burner and chimney design and in 
mantles which have occurred concurrently with the change in size. 
With the coming of this small unit the last excuse for the use 
of open gas jets has become invalid. The only possible remaining 
reason for continuing the use of open jets is the objection to the 
color of the mantle burner, and even this is rapidly disappearing 
in the changes of mantle manufacture, of which mention will be 
made later. The chief fault of the small Welsbach has been the 
clouding of the mica chimney, which depreciates the candle power 
of the burner very materially, while the mantle itself is still in first- 
class condition. This difficulty has been partly at least remedied 
by slight changes in the dimensions of the chimney, and it is not 
of any great importance practically, since the replacement is so 
cheap and the saving of gas so considerable that it is not worth 
while to work the burner beyond the point of good efficiency from 
whatever cause. 


The statements here made with reference to consumption of 
gas per initial candle power are necessarily somewhat loose on 
account of variations in the composition and pressure of the gas 
and in the manufacture of the mantles themselves. The figures 
given are based on a good gas of about 20 so-called candle power 
supplied at the pressure of 2 to 2.5 inches. The efficiency of 
a given gas in a mantle burner bears no definite relation to its 
nominal candle power, nor indeed to its thermal value, so that one 
can deal only in general figures. Variation of an inch of pressure 
either way will affect the candle power in the corresponding direc- 
tion somewhat like 20 per cent, and variations of this magnitude 
are only too common in ordinary gas service. Beside the uncer- 
tainty which may arise from varying quality and pressure of gas, 
it must be remembered that a certain amount of variation must 
be expected in the mantles themselves, particularly if they are 
not from the same batch or from the same manufacturer. Gas 
mantles cannot be readily sorted by actual test burning as are 
incandescent electric lamps, and consequently may at times show 
irregular results. 

The latest forms of mantle gas burner differ somewhat radically 
from those just described, in that they are inverted; that is, the 
Bunsen flame burns upside down and plays upon the interior of a 
round-bottomed bag-shaped mantle. The result of this arrange- 
ment is advantageous, in that the burner thus arranged gives a 
particularly good distribution of light downward, which is often 
desirable, and also a materially higher efficiency than the ordinary 
upright mantle. The gain in efficiency is due chiefly to the fact 
that the products of combustion are carried more effectively away 
from the mantle and the supply of air is brought more advan- 
tageously to it, so that there is a better surface combustion and 
a higher incandescence. 

A good many difficulties had to be overcome in making this 
upside-down burner a success. A Bunsen flame has to be very 
carefully adjusted and regulated in order to burn upside down 
with any degree of stability, and the carrying of the heat into 
and around the burner instead of away from it involves some 
special difficulties. In various ingenious ways these have been 
for the most part overcome, so that the inverted burner is 
gradually displacing the older form. A typical inverted Welsbach 
is shown in cross section in Fig. 49, in which the several parts 


are plainly labeled. The mantle is surrounded by a refractory 
glass chimney with air holes near the lower end for the admission 
of the air supply, which chimney serves to steady the otherwise 
somewhat unstable flame; and commonly an outer globe with a 
hole at the bottom surrounds the chimney proper. To avoid the 
use of wire gauze over the Bunsen tube to prevent flashing back, 

Fig. 49. 

a very ingenious automatic thermostat takes its place in the 
inverted burner. The gauze proved particularly troublesome on 
this type of burner owing to a tendency to clog, while the thermo- 
stat performs the function of the gauze when closed at the first 
lighting of the burner and afterwards opens up, leaving a clear 
passage as the burner becomes thoroughly heated. The mantle 
itself is mounted on a magnesia-clay ring which slips into place 


with a bayonet catch. The inverted mantle can be attached to 
its supporting ring to rather better advantage than the upright 
mantle, and the mass of the mantle itself being smaller, the whole 
affair is less liable to fracture from vibration, so that it has been 
freely used for lighting railway trains with Pintsch gas. 

The typical inverted burner, such as is here described, con- 
sumes about 3 feet of average good gas per hour and gives between 
50 and 55 mean spherical candle power when new, thus doing 
better in the point of economy than the older type of burner. 
Inverted burners of larger capacity with larger mantles have come 
into som<e use for street lighting, and a smaller type corresponding 
somewhat to the Welsbach Junior, just referred to, is being devel- 
oped for interior lighting. The advent of the inverted burner 
makes it possible to use a mantle gas burner in a position which is 
sojnetimes very advantageous and which previously had been the 
sole prerogative of electric incandescent lamps. In this country 
larger units than 75 to 100 candle power are generally secured by a 
duplication of the mantles, forming the so-called gas arc lamps, 
which are used both indoors and out of doors where a light of 
several hundred candle power is desired. Fig. 50 shows a five- 
mantle inverted gas arc intended for outdoor use. This lamp 
utilizes the ordinary inverted mantles, consuming about 3 feet of 
gas each per hour and performing at about the same efficiency as 
the single burners. 

The most striking and sensational improvement made in gas 
practice in recent years has been the use of gas in mantle burners 
supplied under high pressure. The device is an old one, since for 
at least ten years pressure machines of various types have been 
in successful use to a moderate extent for feeding ordinary Wels- 
bachs at enhanced gas pressure. It must not be understood that 
the gas is carried against the mantle at high pressure, but is fed 
into the mixing tube in such a way as to draw in an adequate 
addition of air and thus practically to force the combustion 
within the burner to a point that cannot be reached when sup- 
plying gas at the usual pressure. It is merely a device for secur- 
ing the intense combustion of a considerable volume of gas within 
a very small space, thus forcing the mantles to extremely high 

There have been various modifications of this compressed gas 
supply plan, in some cases the air being artificially compressed; 


and even the use of pure oxygen piped to the gas lamps has 
been tried. Current practice has settled the pressure at 40 to 
60 inches of water. This is obtained by various means. In 

Fig. 50. — Five-mantle Inverted Gas Arc Intended foe Outdoor Use. 

some cases an automatic blower worked by a rudimentary hot-air 
engine supplied from the waste heat of the burners has been tried. 
One form of lamp recently exploited abroad has even gone to the 
length of working a small electric fan at the bottom of the lamp by 



means of current derived from a ring-shaped thermopile operated 
by the waste heat. As a matter of practice, these individual lamp 
compressors have found less favor than the somewhat simpler plan 
of piping the burners for pressure gas supplied from a central 
compressor station. 

In Berlin, which city has the largest example of a press-gas 
plant, several thousand powerful lamps for the streets are sup- 
plied from compressor stations driven by small gas engines and 
located at convenient points, whence the gas is distributed in 
Mannesmann steel tubes, at a pressure of 12 centimeters of mer- 
cury. The press-gas lamps used on this and similar systems have 
one or more large Welsbach mantles, the upright ones being some 
6 inches in height and the inverted ones of correspondingly large 
dimensions. They are woven of extra-heavy material, and though 
owing to the high incandescence the life is much shorter than in 
the low-pressure mantles, it is said to extend to nearly a month 
under ordinary conditions of burning. The ordinary press-gas 
equipments used for street lighting consume from 25 to 35 feet 
of gas per hour, and give, according to pressure, from 30 to 40 
mean spherical candle power per foot of gas. Roughly, there- 
fore, they have more than double the efficiency of the burners 
worked at ordinary pressure, and give the advantage of a simple 
and easily maintained large lighting unit well adapted for outside 

In its evolution press-gas has passed through various stages of 
pressure, the earlier burners being worked at 8 to 12 inches of 
water pressure, but later the more general introduction of pressure 
mains has led to the employment of the higher pressure already 
referred to. Press-gas lighting has not yet been utilized in this 
country to any material extent, although some small experimental 
installations have been set up. The separate piping required 
involves so considerable an expenditure that it has not appealed 
to most gas engineers, and at prices current for gas in America 
the cost of gas, maintenance, and fixed charges have been rather 
too high to permit competition with electric arcs on a consider- 
able scale. The press-gas lamps, however, are most interesting 
examples of high efficiency gas lighting and give an admirably 
steady and powerful light. 

In all incandescent gas lamps the vital point is the mantle, and 
recent improvements have been directed chiefly toward securing 


better material and more uniform methods of manufacture. Prac- 
tically all the ordinary upright mantles manufactured in this 
country at the present time are made of cotton. Cotton fiber 
gives a strong mantle, but one that seems to be more subject 
to deterioration in service than mantles made of some other 
materials. Abroad, the major part of the mantles, and some of 
the inverted mantles in this country are made of ramie fiber. 
This produces a mantle less strong mechanically than a cotton 
mantle, but comparatively more homogeneous in structure arid 
likely to hold up better. 

Within the past two or three years, however, a very radical step 
has been taken in the substitution of artificial silk as a mantle 
material for any of the natural fibers. This artificial silk is simply 
cellulose nitrated in a very moderate degree, substantially in fact 
collodion gun-cotton. This is inflammable without being explosive, 
and in manufacture is squirted into very fine threads possessing 
the luster and almost the flexibility of silk fibers. The material 
thus produced is wonderfully homogeneous, so that mantles made 
from it are exceptionally uniform, while also strong and holding 
up better in candle poyrer than mantles made of natural cellulose. 
Many of these artificial silk mantles are in use on the Continent, 
but as yet they have only been manufactured experimentally in 
this country. The results obtained with them, however, are very 
encouraging, the artificial cellulose fiber having apparently some- 
thing of the same advantage in the incandescent gas mantle that 
it had in the manufacture of the carbon electric incandescent lamp, 
which went through precisely the same steps of evolution, passing 
from one natural fiber to another, and finally settling to the struc- 
tureless artificial cellulose as the most uniform and successful 

Fig. 51 shows in abstract life curves from three types of mantle, 
— inverted mantles of ramie fiber and artificial silk, and upright 
mantles of cotton. It will be noted that the two former hold 
up wonderfully well, while the latter loses about one-third of its 
light in a thousand hours. At first thought, it seems singular that 
the material of the original mantle, which is utterly consumed in 
the process of manufacture, leaving only a skeleton of the oxides 
with which the saturated thread is charged, should make so pro- 
found a difference in the behavior of the mantle; but the finished 
mantle, which is merely the ash, partakes in every particular of the 



structure of the base on which it was formed, and the cotton gives 
a less smooth and dense formation, much more liable to shrinkage 
and breaking away of the finer portions of the structure than 
does the ramie fiber or the artificial silk. 
















VQP 8 











Fig. 51. 


It will be noted that the last two give mantles that hold up 
their candle power almost to the initial point; and in fact in these 
curves, which are laboratory tests of experimental mantles, actually 
ran four thousand hours of total life without material depreciation 


of candle power. It is hardly probable that the general run of 
commercial mantles of this material would turn out as well as 
those here shown, any more than is the case with other com- 
mercial products. The results from the cotton mantle are from 
tests of a considerable group of stock mantles. Nevertheless, 
the probable improvement in the average quality of mantles by 
the use of ramie fiber, and particularly of the artificial silk, is very 

Another point in which large modifications have been made in 
mantle-burner practice is the color. As is familiar to everyone, 
the ordinary mantle of the past has, after a short period of burn- 
ing, tended to pass over from a hue fairly near white to one 
strongly tinged with greenish. A glance at Fig. 46 shows the 
probable cause of this effect. The less ceria the mantle contains 
the more the color of the incandescence tends to greenish or bluish; 
the more ceria, the less the selective radiation in the blue and 
green and the stronger the light in the yellow and orange. The 
ceria apparently tends to burn out of the mantle to a certain extent, 
or at least to lose its activity, so that a mantle starting at one 
per cent of ceria degenerates after a while to the state of a mantle 
containing a considerably less percentage, losing in efficiency and 
acquiring the characteristic greenish tinge. A mantle of composi- 
tion to give it maximum efficiency is, then, liable to this particular 
kind of deterioration. 

Within the past two or three years a good many mantles con- 
taining slightly more ceria have been put out, which give at first 
a light tending more to the yellow, and, while showing some 
selective radiation in the green at a later stage, never during 
their effective life seem to pass to the color of the earlier mantles. 
Still more recently, further advance has been made in this direc- 
tion, and mantles are now available giving a very soft yellowish 
light for a long period of burning. They are, of course, somewhat 
less efficient than the mantles containing less ceria, but are very 
much more agreeable in color and hence are better adapted for 
interior lighting. Mantle manufacture is another of the many 
cases in which the attempt to get the very highest possible effi- 
ciency leads to sacrifice in other directions. These recent types of 
mantle have not come into very great use as yet, but effectively 
meet the requirement of furnishing a light "on the yellow" rather 
than " on the green" in hue. 


Ives (Transactions Illuminating Engineering Society, Vol. V, 
page 208) has given some interesting data on the color variation 
produced by varying amounts of ceria. The upshot of his figures 
is that while a mantle containing three-fourths of one per cent 
of ceria shows rather strong selective radiation in green and bluish 
green, an increase to 1.25 per cent of ceria makes a radical change 
in this particular and gives a fair approximation to the hue of the 
metallic filament electric lamp or the acetylene flame. It is still 
slightly stronger in the green and slightly weaker in the red than 
these, but a small additional amount of ceria has been found to 
even things up in a very satisfactory way. As Fig. 46 shows, the 
loss in efficiency even for as high as two per cent of ceria is 
not at all a serious matter. Some space has been devoted to 
this matter because, while the whole problem has not yet been 
thoroughly worked out, it is clear that the color difficulty found 
with the earlier mantle burners can be and has been overcome to 
a very considerable degree. 

The advantage of the mantle burner in steadiness and economy 
is so great that there would be no reason for using the more 
common forms of gas burner indoors, except for their occasion- 
ally better artistic effects and for their convenience for very 
small lights. The color question and the fragility of the mantle 
have been the chief hindrances to the general introduction of 
the Welsbach type, and these are certainly in large measure 

Recently there have been introduced several forms of mantle 
burner worked with gas generated on the spot from gasoline or 
similar petroleum products. Sometimes these are operated as 
individual lamps and sometimes as small systems to which the 
gas-forming fluid is piped. They give, of course, a fine, brilliant 
light, and at a low cost — cheaper than ordinary mantle burners 
worked with any except rather cheap gas. Where gasoline gas 
would be cheaper than gas taken from the nearest available main, 
such gasoline mantle burners will prove economical. 

But, as a matter of fact, lamps locally generating and burning 
their own petroleum gas have been pretty thoroughly tried from 
time to time during the past twenty-five years, and have never 
taken a strong or permanent hold on the public. It is therefore 
difficult to see how mantle burners worked in similar fashion are 
likely to take a material hold upon the art, although in special 


cases they may prove very useful, when illuminating gas is not 
available at a reasonable price. 

It must be constantly borne in mind that the lighter petroleum 
oils are dangerous and must be used with extreme care, and also 
that they are likely steadily to rise in price, owing to the increasing 
use of explosion engines and gas machines. 

In using any mantle burner it is good economy to replace the 
mantle after three or four hundred hours of burning, if it is in 
regular use to any considerable extent. Of course, in cases when 
a burner is not regularly used and its maximum brilliancy is riot 
at all needed, the mantle may properly be used until it shows 
signs of breaking. In other words, as soon as a mantle which is 
needed at its full efficiency gets dim, throw it promptly away; 
but so long as it gives plenty of light for its situation, your con- 
sumption of gas will not be diminished by a change. 

The commonest trouble with mantles is blackening from a 
deposit of soot owing to temporary derangement of the burner. 
This deposit can generally be burned off by slightly, not consider- 
ably, checking the air supply so as to send up a long, colorless 
flame which will soon get rid of the carbon, after which the full 
air supply should be restored. Too great checking of the air 
supply produces a smoky flame. 

It should finally be noted that the mantle burners are particularly 
useful in cases of troublesome fluctuations in the gas supply, since, 
while they may burn more or less brightly according to circum- 
stances, they are entirely free from rapid flickering when properly 

In leaving now the illuminants which depend upon the com- 
bustion of a gas or liquid, a brief summation of some of their 
properties may not come amiss. 

The replacement of candles and lamps by gas worked a revolu- 
tion, not only in the convenience of artificial lighting, but in its 
hygienic relations. The older illuminants in proportion to their 
luminous effect removed prodigious amounts of oxygen from the 
air and gave off large quantities of carbonic acid. In the days 
of candles a brilliantly lighted room was almost of necessity one 
in which the air was bad. The following table, due to a well- 
known authority on hygiene, gives the approximate properties of 
the common illuminants of combustion as regards their effects on 
the air of the space in which they are burned: 



Tallow candles 2200 grains 

Sperm candles 1 1740 " 

Paraffin oil 992 " 

Kerosene oil. . : I 909 " 

Coal gas, bat wing 5.5 cu. ft. 

Coal gas, Argand [ 4.8 " 

Coal gas, regenerative I 3.2 " 

Coal gas, Welsbach 3.5 " 








i? n 


n (i 


7 fi 


7 II 


n (i 


4 -A 


■>. X 



To this it may be added that acetylene in these relations is about 
on a parity with the Welsbach burner, and that oil lamps other 
than kerosene, burning whale oil, colza oil, etc., would fall in just 
after candles. It is somewhat startling to realize, but very desir- 
able to remember, that a common gas burner will vitiate the air 
of a room as much as four or five persons, in so far, at least, as 
vitiation can be defined by change in the chemical composition 
of the air. 

The introduction of the mantle burner has greatly improved 
gas lighting from the standpoint of the vitiation of the atmos- 
phere, as a glance at the table will show. For equal light the 
mantle burner, compared with gas flames, in virtue of its higher 
efficiency, produces only something like one-fifth the COi and 
moisture per candle power and removes a similarly small propor- 
tion of oxygen. The vitiation of the air with such burners is 
hardly noticeable, unless they are used in considerable numbers 
in a limited space. Now and then one notices it in entering a 
shop brilliantly lighted with mantle burners, but it has disappeared 
as an important consideration under most circumstances. 

In cost also the modern iUuminants have a material advantage. 
In order of diminishing cost the list would run at current American 
prices of materials about as follows: Candles, animal and vege- 
table oils, gas in ordinary burners, kerosene, acetylene, Welsbachs, 
Welsbachs at high pressure. 



At the present time the mainstay of electric illumination is the 
incandescent lamp, in which a filament of high electrical resistance 
is brought to vivid incandescence by the passage of the electric 
current. To prevent the rapid oxidation of the incandescent ma- 
terial at the high temperature employed, the filament is mounted 
in an exhausted glass globe, forming the familiar incandescent 
lamp of commerce. 

The first attempts at incandescent lamps were made with loops 
or spirals of platinum wire heated by the electric current, either 
in the air or in vacuo, but the results were highly unsatisfactory, 
since in the open air the wire soon began to disintegrate, and 
even in the absence of air its life was short. Moreover, the metal 
itself, being produced in very limited quantities, was expensive 
at best, and rose very rapidly in price under a small increase of 
demand. Having a fairly low specific electrical resistance, the 
wire used had either to be very thin, which made it extremely 
fragile, or long, which greatly increased its cost per lamp. 

Following platinum came carbon in the form of slender pencils 
mounted in vacuo. These, however, were of so low resistance 
that the current required to heat them was too great to allow 
of convenient distribution. 

To get a practical lamp it was necessary to use a filament of 
really high resistance, and which was yet strong enough to keep 
down the cost of replacements. 

Without going into the details of the many experiments on 
incandescent lamps, it is sufficient to say that after much labor 
the problem of getting a fairly workable filament was solved 
through the persistent efforts of Edison, Swan, Maxim, Weston, 
and others, about thirty years ago, the modern art dating from 
about 1880. 

All the carbon filaments are based on the carbonization, out 
of contact with air, of thin threads of cellulose — the essential 
constituent of woody fiber. The early work was in the direction 



of carbonizing thread in some form, or even paper, but Edison, 
after an enormous amount of experimenting, settled upon bamboo 
fiber as the most uniform and enduring material, and the Edison 
lamp came to the front commercially. 

In point of fact, it soon became evident that art could produce 
a far more uniform carbon filament than nature has provided, so 
that of late years bamboo, thread, paper, and the rest have been 
abandoned, and all filaments, save those for some special lamps, 
are made from soluble cellulose squirted into threads, hardened, 
carbonized, and " treated." 

Fig. 52 shows a typical modern incandescent lamp. It consists 
essentially of four parts: the base adapted to carry the lamp in 
its socket, the bulb, the filament, and the filament mounting, 
which includes the leading-in wires. In its original form the bulb 
has an opening at each end, one at the base end through which 
the filament and its mounting are put in place, and another in 
the form of a narrow tube a few inches long, which when sealed 
off produces the tip at the end of the bulb. 

The filament is made in slightly different ways in different fac- 
tories, and the exact details of the process, constantly subject to 
slight improvements, are unnecessary here to be described. Sub- 
stantially it is as follows: The basis of operations is the purest 
cellulose convenient to obtain, filter paper and the finest absorbent 
cotton being common starting points. The material is pulped, as 
in paper making, dissolved in some suitable substance, zinc-chloride 
solution being one of those used, evaporated to about the con- 
sistency of thick molasses, and then squirted under air pressure 
into a fine thread, which is received in an alcohol bath to harden it. 

Thus squirted through a die, the filament is of very uniform 
constitution and size, and after carbonization out of contact with 
air it forms a carbon thread that is wonderfully flexible and strong. 
But even so, there is not yet a perfectly uniform filament, and 
the carbon is not dense and homogeneous enough to stand pro- 
tracted incandescence. 

On passage of current portions of the filament mg,y show too low 
resistance, so as to be dull, or too high resistance, so as to get too 
hot and burn off. It is hard, too, to produce a durable filament 
of the somewhat porous carbon obtained in the way described. 

In mn.1ring up the filaments they are, therefore, subjected prior 
to being sealed into the lamp to what is known as the Aching 


process. This has a twofold object, — to build up the filament 
with dense carbon and to correct any lack of uniformity which 
may exist. The latter purpose is far less important to the squirted ■ 

Fig. 52. — Typical Incandescent Lamp. 

filaments than to the old filaments of bamboo fiber or thread, but 
the former is important in securing a uniform product. The fila- 
ments are mounted and then are gradually brought to vivid incan- 
descence in an atmosphere of hydrocarbon vapor, produced from 
gasoline or the like. 


The heated surface decomposes the vapor, and the carbon is 
deposited upon the filament in the form of a smooth, uniform 
coating almost as dense as graphite, and a considerably better 
conductor than the original filament. If, as in the early bamboo 
filaments, there are any spots of poorer conductivity or smaller 
cross section than is proper, these become hot first and are built 
up toward uniformity as the current is gradually raised, so that 
the filament is automatically made unifornl. 

The flashing process is actually quick, the gradual rise of current 
being really measured by seconds. With the squirted filaments 
now used the main value of the flashing process is to enable the 
conductivity of the filament to be quite accurately regulated, at 
the same time giving it a firm, hard coating of carbon that greatly 
increases its durability. The finished filaments are strong and 
elastic, generally a fine steely gray in color, with a polished surface, 
and for lamps of ordinary candle power and voltage vary from 
6 to 12 inches in length, with a diameter of 5 to 10 one-thousandths 
of an inch. 

The filaments are joined near the base of the lamp to two short 
bits of thin platinum wire which are sealed through one end of 
a short piece of glass tube. Sometimes these platinum leading-in 
wires are fastened directly to the ends of the filament and some- 
times to an intermediary terminal of copper wire attached to the 
filament. Within the tube the platinum wires are welded to the 
copper leads which pass down the mounting tube and are attached 
to the base. The filament itself is cemented to its copper or plati- 
num wires by means of a little drop of carbon paste. 

No effective substitute for platinum in sealing through the glass 
has yet been found, although many have been tried. Platinum 
and glass have very nearly the same coefficient of expansion with 
heat, so that the seal remains tight at all temperatures without 
breaking away. It is possible to find alloys with nearly the right 
coefficient of expansion, but they have generally proved unsatis- 
factory either mechanically or electrically, so that the line of 
improvement has mainly been in the direction of making a very 
short seal with platinum wires. 

The filament thus mounted is secured in the bulb by sealing 
the base of the mounting tube or lamp stem into the base of the 
bulb. This leaves the bulb closed except for the exhaustion tube 
at its tip. 


The next step is the exhaustion of the bulb. This used to be 
done almost entirely by mercury pumps, and great pains was taken 
to secure a very high degree of exhaustion. It was soon found 
that there was such a thing as too high exhaustion, but the degree 
found to be commercially desirable is still beyond the easy capa- 
bilities of mechanical air pumps, at least for regular and uniform 
commercial practice, although they have been sometimes success- 
fully used. 

At the present time the slow though effective mercury pump is 
being to a very large extent superseded by the Malignani process, 
or modifications thereof. The bulbs are rapidly exhausted by 
mechanical air pumps, and when these have reached the con- 
venient limit of their action the residual gas is chemically absorbed 
by the vapor produced by heating a small quantity of amorphous 
phosphorus previously placed in a tubulaire connected with the 
exhaustion tube. The process is cheap, rapid, and effective, and 
with a little practice the operator can produce exhaustion that is 
almost absolutely uniform. 

Whatever be the method of exhaustion, during its later stages 
current is put on the filaments both to heat them, and thus to 
drive out the occluded gases, and to serve as an index of the ex- 
haustion. When exhaustion is complete the leading-in tube is 
quickly sealed off, and the lamp is done, save for cementing on 
the base and attaching it to the leads that come from the seal. 
After this the lamps are sorted, tested, and made ready for the 

The shape of the filament in the lamp was originally a simple 
U, later often modified to a U with a quarter twist so that the 
plane of the loop at the top was 90 degrees from its plane at the 
base. As the voltage of distribution has steadily crept upwards 
from 100 to 110, 120, 140, and even 250 volts, it has been necessary 
either to increase the specific resistance of the filament, to decrease 
its diameter, or to increase its length, in order to get the necessary 
resistance to keep the total energy, and likewise the temperature 
of the filament, down to the desired point. 

But the modern flashed filament cannot be greatly increased in 
specific resistance without impairing its stability, so the filaments 
have been growing steadily finer and longer. At present their form 
is various, according to the judgment of the maker in stowing 
away the necessary amount of filament within the bulb. 



One very common form is that of Fig. 52, where the filament has 
a single long convolution anchored to the base at its middle point 
for mechanical steadiness. Sometimes there are two convolutions, 
or even more, and sometimes there is merely a reduplication of 
the old-fashioned simple loop, as in Fig. 53. 

Fig. 53. — Lamp with Double Filament. 

The section of the filaments is now always circular, although 
in the early lamps it was sometimes rectangular or square. 

There has been a considerable fog of mystery about incandescent 
lamp manufacture, for commercial purposes, but the general facts 
are very firmly established and by no means complicated, and a 
little consideration of them will clear up much of the haze. 

To begin with, it is not difficult to make a good filament, but 
it takes much skill and practice to produce, in quantity, one that 


shall be uniformly good. The quality of the lamps as to durability 
and other essentials depends very largely on the care and consci- 
entiousness of the maker in sorting and rating his product. 

It is practically impossible, for example, to make, say, 10,000 
filaments, all of which shall give 15 to 17 horizontal candle power 
at a particular voltage, say 110. With great skill in manufacture, 
half or rather more will fall within these limits, the rest requiring 
anywhere between 100 and 120 volts to give that candle power. 
Only a few will reach these extremes, the rest being clustered more 
or less closely around the central point. 

The value of the lamps as sold depends largely on what is done 
with the varying ones and how carefully they are sorted and rated. 
If the lamps demanded on the market were all of 110 volts, then 
there would be a large by-product which would either have to be 
thrown away, sold for odd lamps of uncertain properties, or slipped 
surreptitiously into lots of standard lamps. 

But some companies use lamps of 108 or 112, or some neighbor- 
ing voltage, and part of the product is therefore exactly fitted to 
their needs, and so forth, there being involved only some slight 
gain or loss in efficiency, not important if similar lamps from other 
lots are conscientiously rated along with them. 

The basic facts in incandescent lamp practice are two: First, 
the efficiency, i.e., the ratio of energy consumed to light given per 
unit of surface, depends mainly on' the temperature to which the 
filament is carried; second, the total light given is directly propor- 
tional to the filament surface which radiates this light. The specific 
radiating power of modern carbon filaments is substantially uni- 
form, so that if one has two filaments of the same surface brought 
to the same temperature of incandescence, they will work at sub- 
stantially the same efficiency and give substantially the same 
amount of light. 

And if a filament of a certain surface be brought to a certain 
temperature, it will give a definite total amount of light, utterly 
irrespective of the form in which the filament is disposed. Changes 
in the form of the filament will produce changes in the distribution 
of the light in different directions around the lamp, but will not 
in the least change the total luminous radiation. Much of the 
current misunderstanding is due to neglect of this simple fact. 

The nominal candle power of the lamp depends upon a pure 
convention as to the direction and manner in which the light shall 



be measured in rating the lamp, and makers have often sought to' 
beat the game by disposing the filament so as to exaggerate the 
radiation in the conventional direction of measurement. 

For example: Many early incandescent lamps had filaments of 
square cross section bent into a single simple U. These gave their 
rated candle power in directions horizontally 45 degrees from the 
plane of the filaments, and this was the maximum in any direction, 
so that the lamp when thus measured was really credited with its 
maximum candle power, and fell below its rating in all directions 
save the four horizontal directions just noted. 

Fig. 54. — Distribution of Light from Flat Filament. 

It is customary to delineate the light from an incandescent lamp 
in the form of closed curves, of which the various radii represent 
in direction and length the relative candle power in those various 
directions. Such curves may be made to show accurately the dis- 
tribution of light in a horizontal plane about the lamp, or the 
distribution in any vertical plane, and from the average radii in 
any plane may be deduced the mean candle power in that plane, 
while from a combination of the radii in the various planes may 
be obtained the mean spherical candle power which measures the 
total luminous radiation in all directions. 

This last is the true measure of the total light-giving power of a 
lamp. Fig. 54 illustrates the curve of horizontal distribution for 



one of the early lamps, having a flat U-shaped filament. The 
circle is drawn to show a uniform 16 candle power, while the 
irregular curve shows the actual horizontal distribution of light. 
This particular lamp overran its rating, but its main characteristic 
is that it gave a strong light in one horizontal diameter and a 
weak one in the diameter at right angles to this. 

Such a distribution as this is generally objectionable, and most 
modern filaments are twisted or looped, so that the horizontal 
distribution is nearly circular. Fig. 55 shows a similar curve for 
a recent 16-c.p. lamp of the type shown in Fig. 52. In the small 
inner circle is shown the projection of the looped filament as one 
looks down upon the top of the lamp. Fig. 56 shows a similar 

Horizontal Distribution 

Vertical on 90 Horizontal 

Figs. 55 and 56. — Distribution of Light from Looped Filament. 

delineation of the distribution of light in a vertical plane taken 
in the azimuth shown in Fig. 55, with the socket up. 

The looping of the filament is such that the horizontal distri- 
bution is very uniform, while in the vertical downwards there is a 
marked diminution of light, and of course in the direction of the 
socket most of the light is cut off. The total spherical distribu- 
tion, if one can conceive it laid out in space in three dimensions, 
resembles a very flat apple with a marked depression at the blossom 
end and a cusp clear into the center at the stem end. Fig. 57 
is an attempt to display this spherical distribution to the eye. 

If the filament were a simple U, or the double U of Fig. 53, 
assuming the same total length and temperature of filament, the 
apple would have still greater diameter, but the depression at 
the blossom end would be considerably wider and deeper. 


If the filament has several convolutions, as in Fig. 58, this 
depression is considerably reduced, but there is a marked flatten- 
ing in one horizontal direction, so that the horizontal distribu- 
tion' would somewhat resemble Fig. 54. But the total luminous 
radiation would be quite unchanged. 

If the lamps were rated by their mean horizontal candle power, 
the U filament would show abnormally large horizontal illumi- 
nation for the energy consumed, and would apparently be very 
efficient, while if one were foolish enough to rate lamps by the 

Fig. 57. — Distribution of Light from Incandescent Lamp. 

light given off the tip alone, Fig. 58 would show great efficiency, 
the distribution in one horizontal diameter having been reduced 
to fatten the curve at the tip. In reality, however, each one of 
the three forms of lamp would have exactly the same efficiency, 
and in practice there would be little choice between them. 

In the everyday work of illumination carbon incandescent lamps 
are installed with their axes in every possible direction, the vertical 
being the rarest, and angles between 30 degrees and 60 degrees 
downwards from the horizontal the commonest. 

Bearing in mind this general distribution of the axes and the 
fact that diffusion goes very far toward obliterating differences in 



the spherical distribution as regards general illumination, it is 
easy to see that the shape of the filament is, for practical purposes 
of illumination, of little account. In 
the few cases where directed illumi- 
nation is needed it is best secured 
by a proper reflector, which gives far 
better results than can be obtained 
by juggling with the shape of the 

The thing of importance is to get 
uniform filaments of first-class dura- 
bility, and of as good efficiency as 
possible. . The only proper test for 
efficiency, however, is that based on 
mean spherical candle power, since 
a lamp will give a different apparent 
efficiency for each direction of meas- 
urement, varying from zero in the 
direction of the socket to a maxi- 
mum in some direction unknown until 
Efficiency has most often been taken 
™ — ■ with respect to the mean horizontal 

8. -Lamp with Multiple candle power , But thifi j^^ to cor _ 
looped Filament. 

Fig. £ 

rect relative results only when compar- 
ing lamps having filaments similarly curved. The mean' spherical 
candle power is usually from 80 to 85 per cent of the mean hori- 
zontal candle power. 

As regards efficiency, most commercial incandescent lamps re- 
quire between 3 and 4 watts per mean horizontal candle power. 
Now and then lamps are worked at 2.5 watte per candle when 
used with storage batteries, and some special lamps, especially 
some of those made for voltages above 200, range over 4 watts 
per candle. As has already been remarked, the efficiency depends 
upon the temperature at which the carbon filament is worked. 
And it is in the ability to stand protracted high temperature that 
filaments vary most. 

It is comparatively easy to make a filament which will stand 
up well when worked at 4 watts per candle, but to make a good 
3-watt-per-candle filament is a very different proposition. Also, 



at low voltage, 50 volts for instance, the filament is more sub- 
stantial than the far slenderer one necessary to give the requisite 
resistance for use at the same candle power at 100 or 125 volts. 

Under protracted use the filament loses substance by slow disin- 
tegration and by a process akin to evaporation, so that the surface 
changes its appearance, the resistance increases so that less current 
flows, the efficiency consequently falls off, and the globe shows 
more or less blackening from an internal deposit of carbon. 

The thinner and hotter the filament the less its endurance and 
the sooner it deteriorates or actually breaks down. Modern car- 
bon lamps have by improved methods of manufacture been devel- 
oped to a point that in the early days of incandescent lighting 
would have seemed beyond hope of reach. But the working 
voltage has steadily risen and constantly increased the difficulties 
of the manufacturer. 

So-called high-efficiency lamps worked at about 3 watts per 
candle power require the temperature of the filament to be carried 
so high that its life is seriously endangered unless it be of fair 
diameter; hence such lamps are hard to make for low candle 
power or for high voltage, either of which conditions requires a 
slender filament — in the former case to limit the radiant surface, 
in the latter to get in the needful resistance. An 8-c.p. 125-volt 
lamp, or a 16-c.p. 250-volt lamp, presents serious difficulties if the 
efficiency must be high, while lamps of 24 or 32 candle power are 
far more easily made for high voltage. 

The annexed table gives a clear idea of the performance of a 
carbon filament lamp under various conditions of working. It is 
from tests made on a 16-c.p. 100-volt lamp (so-called) by Prof. 
H. J. Weber. 





Watts per 
Candle Power. 







1464° C. 
























































The absolute values of the temperatures here given are the 
least exact part of the table, but the relative values may be 
trusted to a close approximation. More recent data indicate that 
the true filament temperatures range from about 1800° C. in a 
4-watt-per-candle lamp to nearly 1950 degrees at 3.1 watte per 
candle. Fig. 59 shows in graphical form the relation between the 
last two columns, showing clearly how conspicuously the efficiency 
rises with the temperature. At the upper limit given the carbon 
is too hot to give a long life, although the writer has seen modern 
lamps worked 12 volts above their rating for several hundred 
hours before rapid breakage began. Of course the brilliancy had 
fallen off greatly by that time. 

Watts per mean horizontal candle power 
Fig. 59. — Variation of Efficiency with Temperature. 

It is worth noting from the table that for a 16-c.p. lamp of 
ordinary voltage the candle power varies to the extent of quite 
nearly one candle power per volt, for moderate changes of voltage 
from the normal. Weber calls attention to the fact that between 
1400 degrees and 1650 degrees an increase in temperature of n 
degrees corresponds very closely to a saving in energy of n per cent 
in the production of light. 

If it were possible to cany the temperature still higher without 
seriously impairing the stability of the filament, lamps of a very 
high economy could be produced. It is possible to force lamps 
up to an economy of even 1.5 watts per candle temporarily, but 
they often break almost at once, and even if they hold together 
they rise to 2 or 2.5 watts per candle within a few hours. 

To tell the truth, the temperature corresponding to 1.5 watts per 
candle is dangerously near the vaporizing point of the material, — 
so near that it is practically hopeless to expect any approximation 
to such efficiency from carbon filaments, and even at 2.5 watte per 


candle the life of the lamps is so short that at present prices they 
cannot be used commercially. 

From such experiments as those tabulated it has been shown 
that the relation between the luminous intensity and the energy 
expended in an incandescent lamp may be expressed quite nearly 
by the following formula: 

/ = aTP, 

wherein/ is the candle power, W the watts used, and a is a quantity 
approximately constant for a given type of lamp, but varying 
slightly from type to type.. 

Following the universal rule of incandescent bodies, the radia- 
tion from an incandescent lamp varies in color with the tempera- 
ture, and thus as the voltage changes, or what is about the same 
thing, as lamps of different efficiencies are used, the color of the 
light varies very conspicuously. Low efficiency lamps, or lamps in 
a low stage of incandescence, such as is indicated in the first four 
lines of the table, burn distinctly red or reddish orange. Then 
the incandescence passes through the various stages of orange- 
yellow and yellow until a 3-watt lamp is clear yellowish white and 
a 2.5-watt lamp still more whitish. The color is a good index of 
the efficiency. 

The sizes of carbon incandescent lamps in common use are 8, 10, 
16, 20, 24, and 32 candle power. The standard in this country is 
the 16-c.p. size, a figure borrowed from the legal requirements for 
gas. Some 10 candle power lamps are used here, very few 8 candle 
power, and still fewer of candle powers above 16. Abroad, 8-c.p. 
lamps are used in great numbers and with excellent results. The 
20-c.p. and 24-c.p. lamps are found mostly in high voltages, for 
reasons that will appear shortly. Two-, 4-, and 6-c.p. lamps are 
considerably used for decorative purposes or for night-lights, and 
excellent 50-c.p. lamps are available for cases requiring radiants of 
unusual power. 

Lamps of these various sizes are made usually for voltages be- 
tween 100 and 120 volts, and more rarely for 220 to 250 volts, but 
in the latter case lamps below 16 candle power are used in America 
only to a very small extent. 

In lamps of small candle power or. of high voltage there is 
some temptation to get resistance by flashing the filaments less 
thoroughly, to the detriment of durability, since the soft core dis- 


integrates more readily than the hard deposited carbon, which may 
explain the frequent inferiority of such lamps. The greater the 
candle power, and the less efficiency required, i.e., the greater the 
permissible radiating surface, the easier it is to get a strong and 
durable filament for high voltages. Hence, lamps for 220 to 250 
yolts are generally of at least 16 candle power, very often of 20 or 
24 candle power, and seldom show an efficiency better than 4 watts 
per candle power. 

This forms a serious practical objection to the use of such lamps 
for general distribution, unless with cheap water power as the source 
of energy, and while improved methods of manufacture are likely 
somewhat to better these conditions, yet there are inherent reasons 
why it should be materially easier to produce durable and efficient 
incandescent lamps of moderate candle power and voltage than 
lamps of extreme properties in either of these directions. 

If the lamp is started at a low efficiency, the temperature is 
relatively low and the decadence of the filament is retarded, while 
if the lamp is initially of high efficiency the filament under the 
higher temperature deteriorates more rapidly and the useful life of 
the lamp is shortened. 

Under this latter condition the cost of energy to run the lamp 
is diminished, but at the price of increased expense in lamp re- 
newals. Operating at low efficiency means considerable cost for 
energy and low cost of the lamp renewals. Between these diver- 
gent factors an economic balance has to be struck. 

It is neither desirable nor economical to operate an incandescent 
lamp too long, since not only does it decrease greatly in efficiency, 
but the actual light is so dimmed that the service becomes poor. 
If the lighting of a room is planned for the use of 16-c.p. lamps, 
and they are used until the candle power falls to, say, 10, which 
would be in about 600 hours in an ordinary 3-watt-per-candle 
lamp, the resulting illumination would be altogether unsatisfactory. 
Quite aside from any consideration of efficiency, therefore, it be- 
comes desirable to throw away lamps of which the candle power 
has fallen below a certain point. 

Much of the skill in modern lamp manufacture is directed to 
securing the best possible balance between efficiency and useful 
life, a thing requiring the most painstaking efforts of the manu- 
facturer. Fig. 60 shows graphically the relation between life, 
candle power, and watts per candle derived from tests of high- 


grade foreign carbon lamps. In comparing these, like the previous 
data, with American results, it should be borne in mind that these 
foreign tests are made, not in terms of the English standard candle, 
but generally in terms of the Hefner-Alteneck standard, which is 
somewhat (approximately 10 per cent) smaller. 

These curves show the results from lamps having an initial 
efficiency of 2.5, 3.0, and 3.5 watts per candle power and an initial 
candle power of 16. They show plainly the effect of increased 
temperature on the life of the lamp, and it is unpleasantly evident 
that in the neighborhood of 3 watts per candle a point is reached 
at which a further increase of efficiency produces a disastrous result 

Curves a-Wntts per C.P. Curves b -CP. 

Fig. 60. — Curves Showing Life, Candle Power, and Watte per Candle. 

upon the life; in other words, such efficiency requires a tempera- 
ture at which the carbon filament rapidly breaks down. 

And so long as carbon is used as the radiant material there is a 
strong probability that there can be no very radical improvement 
in efficiency. Of course, if incandescent lamps were greatly cheap- 
ened, it would pay to burn them at higher efficiency and to 
replace them oftener. 

In production on a large scale the mere manufacture of the 
lamps can be done very cheaply, probably at a cost not exceeding 
7 to 8 cents, but the cost of proper sorting and testing to turn 
out a uniform high-grade lamp, and the incidental losses from 
breakage and from lamps of odd and unsalable voltages, raise the 
total cost of production very materially. Much of the reduction 
in the price of incandescent lamps in the past few years has resulted 


from better conditions in these latter respects, as well as from the 
improved methods of manufacture. 

And it should be pointed out that the difference between good 
and bad lamps, as practically found upon the market, lies mostly 
in their different rates of decay of light and efficiency. It is the 
practice of many of the large lighting companies who renew the 
lamps for which they furnish current to reject and replace lamps 
which have fallen to about 80 per cent of their initial power. 

First-class modern lamps worked in the vicinity of 3 watts per 
candle power will hold up for 400 to 450 hours before falling below 
this limit, and at 3.5 or 3.6 watts per candle power will endure 
nearly double that time. They are often rated in candle-hours 
of effective life, and on the showing just noted the recent high- 
efficiency lamp will give a useful life of 6000 to 7000 candle-hours, 
with an average economy of perhaps 3.25 watts per candle. A 
medium-grade lamp of similar nominal efficiency may not show 
with a similar consumption of energy more than 250 or 300 hours 
of effective life — say 4000 to 4500 candle-hours. 

The economics of the matter appear as follows: The first lamp 
during its useful life of, say, 6500 candle-hours, will consume 21.125 
kilowatt hours, costing at, say, 15 cents per kilowatt hour, $3.17, 
and adding the lamp at 18 cents, the total cost is $3.35, or 0.0515 
cent per candle-hour, while the poorer lamp at 4000 candle-hours 
will use $1.95 worth of energy, and at 18 cents for the lamp 
would cost 0.0532 cent per candle-hour. To bring the two lamps 
to equality of total cost, irrespective of the labor of renewals, 
the poorer one would have to be purchased at 11 cents. In other 
words, poor lamps, if discarded when they should be, generally so 
increase the cost of renewals that it does not pay to use them 
at any price at which they can be purchased under ordinary 

As has already been explained, lamps deteriorate very rapidly 
if exposed to abnormal voltage, and the higher the temperature 
at which the lamp is normally worked the more deadly is the 
effect of increased voltage. It thus comes about that if high- 
efficiency lamps are to be used, very good regulation is necessary. 
Occasional exposure to a 5 per cent increase of voltage may easily 
halve the useful life of a lamp, while, of course, permanent work- 
ing at such an increase would play havoc with the life, cutting it 
down to 20 per cent or less of the normal. Good regulation is, 


therefore, of very great importance in incandescent lighting, not 
only to save the lamps and to improve the service, but to render 
feasible the use of high-efficiency lamps. On the whole, the best 
average results seem to be obtained in working lamps at 3 to 3.5 
watts per candle. Those of higher efficiency fail so rapidly that 
it only pays to use them when energy is very expensive and must 
be economized to the utmost. The 2.5-watt lamp of Fig. 60, for 
example, has an effective life of not more than 150 hours, at an 
average efficiency of about 2.75 watts per candle. A 2-watt lamp 
will fall to 80 per cent of its original candle power in not far 
from 30 hours, at an average efficiency of about 2.25 watts,' while 
if started as a 1.5-watt lamp, in a few hours the, filament is reduced 
to practical uselessness. 

There is seldom any occasion to use lamps requiring more 
than 3.5 watts per candle power, save in case of very high voltage 
iqptallations, where the saving in cost of distribution may offset 
the cost of the added energy. The difficulty of making durable 
250-volt lamps on account of the extreme thinness of the filament 
has been already referred to, and it is certainly advisable to use 
in such installations lamps of 20 candle power or more whenever 
possible, thus making it practicable to work at better efficiency 
without increased risk of breakage. Even when power is very 
cheap there is no object in wasting it, and a little care will gener- 
ally secure regulation good enough to justify the employment of 
incandescent lamps of good efficiency. 

Further, in the commercial use of lamps it is necessary for 
economy that the product should be uniform. It has already 
been shown that medium-grade lamps are characterized by a 
shorter useful life than first-class lamps. Unfortunately, there 
are on the market much worse lamps than those described. It 
is not difficult to find lamps in quantity that are so poor as to 
fall to 80 per cent of their initial power in less than 100 hours. 
A brief computation of the cost of replacement will show that 
these are dear at any price. Now, if lamps are not carefully 
sorted, a given lot will contain both good lamps and poor lamps, 
and will not only show a decreased average value, but will contain 
many individual lamps so bad as to give very poor and uneco- 
nomical service. Fig. 61 shows what is sometimes known as a 
" shotgun diagram," illustrating the variations found in carelessly 
sorted commercial lamps. In this case the specifications called 


for 16-c.p., 3.5-watt-per-candle lamps. The variation permitted 
was from 14.5 to 17.5 mean horizontal candle power, and from 
53 to 59 total watts, which is a liberal allowance, some companies 
demanding a decidedly closer adherence to the specified limits. 

The area defined by these limits is marked off in the cut, form- 
ing the central "target." The real measurements of the lamps 
tested are then plotted on the diagram and the briefest inspection 

Kg. 61. — Shotgun Diagram. 

shows the results. In this case only 46 per cent of the lamps hit 
the specifications. All lamps above the upper slanting line are 
below 3.1 watts per candle power, and hence are likely to give 
trouble by falling rapidly in brilliancy and breaking early. Lamps 
below the lower slanting line are over 4 watts per candle power, 
hence are undesirably inefficient. Moreover, the initial candle 
power of the lot varies from 12.2 candle power to 20.4 candle 


Such a lot will necessarily give poorer service and less satis- 
factory life, and is, as a matter of dollars and cents, worth much 
less to the user than if the lamps had been properly sorted at 
the factory. Filaments cannot be made exactly alike, and the 
manufacturer has to rely upon intelligent sorting to make use of 
the product. For example, the topmost lamp of Fig. 61 should 
have been marked for a lower voltage, at which it would have 
done well. Nearly all the lot would have properly fallen within 
commercial specifications for 16-c.p. lamps at some practicable 
voltage and rating in watts per candle power. The imperfect 
sorting has misplaced many of the lamps and depreciated the 
whole lot. 

In commercial practice lamps should be carefully sorted to 
meet the required specifications, and the persons who buy lamps 
should insist upon rigid adherence to the specifications, and 
should, in buying large quantities, test them to insure their cor- 
rectness. To sum up, it pays to use good lamps of as high 
efficiency as is compatible with proper life, and to see that one 
gets them. 

The real efficiency of an incandescent lamp, i.e., the proportion 
of the total energy supplied which appears as visible luminous 
energy, is very small, ordinarily from 2 to 3 per cent in carbon 
lamps, not over 5 to 6 per cent even in the best metallic filament 
lamps. This means that in working incandescent lamps from 
steam-driven plants less than 0.5 per cent of the energy of the coal 
appears as useful light. 

Up to about 1905 the carbon lamp substantially as just de- 
scribed was the only form of incandescent lamp used in this coun- 
try. At about this time a radical modification in carbon filaments 
was produced which has come into large commercial use. This 
was the so-called metallized filament, substantially an allotropic 
form of carbon which was the result of attempts to convert an 
ordinary carbon filament into pure graphitic carbon. The manu- 
facture of the metallized filament starts with the ordinary squirted 
cellulose filament already described as the base of operations. 
This is baked in an electric furnace at a very high temperature, 
and after subjection to a flashing process akin to that in general 
use is again fired, the temperature being carried to the neigh- 
borhood of 3000° C. The result is a complete change in the 
texture and appearance of the carbon and in its physical proper- 


ties. The specific resistance is enormously reduced, falling to a 
figure comparable with the poorly conducting metals, and the tem- 
perature coefficient becomes positive like that of a metal, instead 
of being negative, as in the ordinary forms of carbon. The metal- 
lized filament in practice has only a very small temperature coeffi- 
cient of either sign, and is, from its lower resistance, much slenderer 
than the ordinary carbon filament and considerably more refrac- 
tory, so that it can be worked at a higher temperature. 

The normal initial specific consumption of these lamps is about 
2.5 watts per m. h. c. p., and the life and fall of candle power 
during life approximately the same as for the 3.1-watt ordinary 
carbon-filament lamps. Such lamps are manufactured in sizes of 
50, 100, 125, 187, and 250 watts, rated respectively at 20, 40, 50, 
75, and 100 candle power. The smallest size has replaced the ordi- 
nary 50-watt 16-c.p. carbon lamp to a very large extent, and the 
larger sizes have been considerably used in commercial lighting, but 
are now rapidly disappearing under the competition of the true 
metallic filament lamps. 

The metallized filament, interesting as it is, was an improvement 
introduced a few years too late, since at the time of its production 
the true metallic filament lamps, now coming into general use, had 
already been produced abroad. The metallized filament lamps, 
therefore, which resemble in general properties and distribution of 
light the carbon lamps which had preceded them, are now only of 
passing interest. The first of the metallic filament group of lamps 
was the osmium lamp of Dr. Auer von Welsbach. Osmium is a 
rare metal found associated with the platinum group in small 
quantities. It has an atomic weight of 191, a specific gravity of 
22.48, and a specific resistance in the lamp filaments of about 47 
microhms per cubic centimeter. It is a strong acid-forming ele- 
ment of extremely uncompromising mechanical qualities, and has 
not yet been produced in true metallic form, but only as a black 
powder. The osmium filaments were made by mixing the finely 
divided metal with a binding material into a paste, squirting it 
through dies into a filament, and then driving out the binder by 
intense heat and sintering the residuum into a coherent metallic 
mass. The relatively high conductivity of the metal forbade the 
successful production of filaments for ordinary voltages, and most 
of the commercial lamps were intended to be burned at a pressure 
of about 50 volts, either on a separate circuit, or two in series on 


the ordinary voltages. Even so, the ordinary lamp of about 20 
candle power contained 3 loops in series, each about 2.5 inches long 
and anchored near the tip of the lamp. The filaments when hot 
were so plastic that the lamp had to be burned tip down. 

The specific consumption of the osmium lamp, however, was so 
low, from 1 to 1.5 watts per candle, that in spite of the difficulties 
of fragility and high cost it came into commercial use on a modest 
scale, and its success was the immediate cause of the further 
researches which led to the development of the metallic filament 
lamps now in common use. The most serious difficulty with the 
osmium as a material for filaments, however, was the extreme rarity 
of the metal, so that if it had been employed to any considerable 
extent the price must inevitably have risen very seriously. As it 
was, the difficulty was felt to an extent which was met by putting 
out the lamps on a nominal lease so as to insure their return and 
the saving of the material. The osmium lamp is now only of his- 
torical interest, but its production was of the greatest importance 
to the industry in stimulating further improvements. 

The first really successful metallic filament lamp in a large com- 
mercial way was the tantalum lamp now used in large quantities. 
Tantalum is another of the relatively rare 
metals, of atomic weight 183, density about 
16.8, and specific resistance about 16.5 mi- 
crohms per cubic centimeter. As prepared 
in the electric furnace, it is a whitish, in- 
tensely hard metal, with about the strength 
of steel, not attacked by any of the ordinary 
acids, and with a melting point considerably 
higher than that of platinum. It is suffi- 
ciently ductile to be drawn into very fine 
wire, which for use in lamps is commonly 
about 2 mils in diameter, of which about 
2 feet is necessary to produce a commercial 
110-volt lamp of 20 to 25 candle power. 

On account of the great length of filament 
to be supported, it has to be strung on sup- 
porting spiders in 10 or more short loops. Rb-62. — Tantalum 
Fig. 62 shows the ordinary commercial tan- p ' 

talum lamp as manufactured in this country. The tantalum lamp 
is ordinarily worked at a specific efficiency of 2 watts per candle. 


At this efficiency its normal life before falling off, say 20 per cent 
in candle power, is in the neighborhood of 800 hours, although 
individual lamps will often burn considerably longer than this with- 
out reaching the loss specified. After protracted use the filament 
tends to draw tight over the spiders and eventually breaks. The 
filament, however, possesses a curious capacity for easy welding, 
so that a broken lamp may often be connected with the circuit and 
manipulated so as to weld a broken length of filament to the next 
stretch, sometimes with a very trivial loss in length, after which 
the lamp may burn for several hundred hours more, and perhaps 
be again rewelded if one follows the operation to its limit. 

The most serious fault of the tantalum filament is its inability 
to give good results on alternating-current circuits. When burned 
upon these the filament tends to break crosswise and weld itself 
together again without actually separating, so that when examined 
under the microscope the filament looks as though it had been many 
times broken and carelessly glued together again. Fig. 63 shows a 
photomicrograph of a filament thus affected. After a couple of hun- 
dred hours burning on an ordinary 60-cycle circuit this " faulting" 
has occurred to such an extent that the filament is very fragile. 
Its life on such a circuit is approximately half that on direct cur- 
rent, although individual lamps may occasionally last 1000 hours 
or more. The cause of this " faulting' ' has never been ferreted 
out, nor has it as yet been remedied. The beautiful mechanical 
properties of the tantalum wire, however, enable commercial lamps 
to be made for as low as 25 watts consumption on 110-volt circuits, 
"and abroad many 220-volt tantalum lamps are in use, two of the 
ordinary spiders being mounted in tandem in a single bulb. The 
cost of the tantalum lamp is approximately double that of the 
carbon lamp, but the very greatly increased efficiency makes its 
use desirable at all ordinary prices for current. It would in fact 
have come into use in enormous quantities had it not been for 
the subsequent production of the tungsten metallic filament lamp, 
which permits a still higher efficiency and is equally available on 
direct- and alternating-current circuits. 

Shortly after the commercial appearance of the tantalum lamp 
experimental work on a tungsten filament was brought to a suc- 
cessful issue, and this metal, owing to its very refractory character, 
can be worked at a higher temperature than any filament yet 
found. Tungsten, although a comparatively rare metal, is much 


more available in quantity than either osmium or tantalum. Its 
atomic weight is 184, its specific gravity about that of platinum, 
and its melting point somewhat in excess of 3000° C. Only at 
the highest temperatures of the electric furnace can tungsten be 
reduced to a state of a coherent metal. And hence most of the 

Kg. ea 

tungsten filaments have been prepared by a method analogous 
to that employed with the osmium filament; that is, the filament 
is made from a mixture or alloy of finely divided tungsten from 
which all other materials other than the tungsten are driven off 
by heat, and the filament remaining is then sintered into coherent 
structure by intense heat. 


At least half a dozen forma of this general process have been 
employed, some of them involving purely mechanical mixtures like 
that used for the osmium filament, and others the use of com- 
pounds or amalgams reduced finally to metallic tungsten and 
sintered. These processes are partially kept secret and often sub- 
ject to change, but they lead to the same final product, — a tung- 
sten filament not like a drawn wire, but a sintered mass more or 
less dense and coherent and giving a workable although somewhat 
fragile filament. Owing to its enormously high melting point, the 
tungsten filaments can be safely worked at a temperature of 
something like 2300° C, at which the specific consumption is 
about 1.25 watts per horizontal candle power, approximately 
1.5 to 1.6 watts per m. s. c. p. As iff the case of other metallic 
filaments, the specific resistance of the 
metal is too low to give a conveniently 
short filament, so that it is necessary, 
as with osmium and tantalum, to use 
several loops in series to obtain a lamp 
which will burn upon the customary 
voltages. The ordinary 25-watt tung- 
sten lamp of commerce usually has 
four such loops, carrying in all 16 to 20 
inches of filament of a diameter slightly 
less than 0.002 of an inch. The ap- 
~~ psj, 64 pearance of the typical lamp is shown 

in Fig. 64. The filaments are usually 
carried as shown, on a long and rather slender spider, so that the 
loops are sufficiently anchored at each end. The average life of 
such tungsten lamps worked at an initial consumption of 1.25 
watts per candle under constant pressure runs to about 1000 
hours, while falling off approximately 20 per cent in candle power. 
Owing to the difficulties of manufacture, great uniformity is much 
more difficult to secure than with the carbon and tantalum fila- 
ments, and consequently a good many lamps fail by breakage 
after a less period of life than this; while others may run on to 
2000 hours or more, leaving the average about as stated. Since 
tungsten, like other metals, has a positive temperature coefficient, 
the variation of candle power and efficiency with voltage is mate- 
rially less than with carbon lamps. Fig. 65 shows for a group of 
25-watt tungsten lamps the relation between watts per candle, 


m. a. c. p., and voltage. The customary working temperature 
of a tungsten filament seems to be fairly near the limit of its 
economical endurance, as in the case of the working temperatures 
as determined by experience for other filaments, so that the increase 
of life with a small drop in voltage or the decrease of life with 
a small increase of voltage is conspicuous. 

■ I 


Fig. 65. 

Fig. 66 shows the relation of life to voltage as determined for 
present commercial lamps having plain carbon, "metallized" car- 
bon, tantalum, and tungsten filaments. It will be noted that in 
the case of the tungsten a drop of 2 per cent in voltage will 
increase the life by about 30 per cent. Tungsten lamps are 
regularly manufactured for the usual voltages of 110 to 120 in 



sizes of 25, 40, 60, 100, 125, 150, and 250 watts; while lamps on 
the one hand of 15 and 20 watts, and on the other of 300, 400, 
and 500 watts, have been experimentally produced and are begin- 
ning to come into use. All are worked at substantially the same 
efficiency, save for the lamps designed, as are some of the larger 
sizes, for 220 to 240 volts, which are rated at about 1.4 watts per 
horizontal candle power. In addition, series incandescent lamps 
of various sizes from 40 to 250 watts for all the usual constant 
currents are manufactured, and these have practically driven the 
carbon series lamps from the field. 





















— Gun. 



- Ti m»i 







x \ 










: 5> 













% 1 










Fig. 66. 

The weak point of the tungsten filament has been its fragility 
owing to the extreme slenderness of the filaments in the smaller 
lamps and the brittleness of the material. When hot the fila- 
ments are amply strong, and it is therefore desirable to turn on 
the lights in case the globes or shades are to be cleaned. The 
earlier tungsten lamps were in fact so fragile that shipment was 
accomplished only with great difficulty and with considerable 
breakage. At present the filaments, while still delicate, are much 
stronger than the earlier ones, and can readily be burned with 


the lamp in any position, a procedure that was distinctly unwise 
at first. 

The changes going on in the production of these tungsten lamps 
are so rapid that it is futile to attempt more than a cursory 
description of the state of the art. The lamps are now manu- 
factured by a great number of makers both here and abroad 
under a wide variety of trade names, but the differences between 
them are mostly of a minor character. The most important recent 
improvement in manufacture, not yet generally introduced, is the 
production of a tungsten filament of drawn wire like the tantalum 
filament. At the extreme temperatures of the electric furnace, 
tungsten can be produced in true metallic state, and although 
brittle when cold it is moderately ductile when hot, and is drawn 
into wire in this state. The details of this difficult process have 
now been fairly well worked out, so that the drawn filaments have 
already come into some commercial use and show evidence of 
considerably greater strength and rather better life than the sin- 
tered filaments previously used. It 'is too early yet to speak 
of their exact properties. The wire drawn filaments are mounted 
on spiders akin to those used in the tantalum lamps. 

Tungsten lamps have already gone far in driving out the carbon 
lamps, and prove particularly useful in the larger units, from 
40 to 150 watts. The only objection to their employment is 
their cost and the fact that they are not yet strong enough to 
stand use in positions where considerable vibration is experienced. 
Both these difficulties are likely to be remedied to a very material 
extent. The present cost of tungsten lamps, being at retail from 
60 cents upwards, according to size, is so considerable that break- 
age becomes a serious item, which is, however, more than over- 
balanced by the saving in current, except where current is to be 
had at very low rates. 

Whether the tungsten lamp represents the last stage of improve- 
ment of the metallic filament remains to be seen. There is cer- 
tainly no metal yet known which has a higher melting point, nor 
is there a reasonable expectation of finding such a metal. It is 
possible that the use of alloys may lead to improvement, but it is 
fairly certain that no alloy is likely to raise the melting point 
already available with tungsten, although it is highly probable 
that it may be possible to work out alloys of the extremely 
refractory metals which will improve their mechanical properties 


and also raise their specific resistance, an end which is highly 
desirable. Many experiments have been tried and are being tried 
with nonmetallic filaments composed of refractory oxides and 
other compounds. None of these has yet reached a point where 
it is more than experimental, qxcept for the Nernst lamp, about 
to be described, which belongs to a somewhat different class. It is 
worth mentioning, as a matter chiefly of theoretical interest, that 
practically all the modern filaments have a somewhat favorable 
selective radiation as compared with carbon. In other words, 
their radiation is more efficient than the true temperature of 
the filament would indicate, owing to a better distribution of 
energy in the spectrum. 

The Nernst lamp, introduced about ten years ago, differs materi- 
ally from the other incandescent lamps in that the material of the 
light-giving body is a nonconductor when cold and has to be oper- 
ated in air rather than in vacuo, to the material disadvantage of 
its efficiency. It was, however, the first incandescent lamp of high 
efficiency to appear, and has found its way into considerable use 
both here and abroad. The basic fact taken advantage of by Dr. 
Nernst in the production of his lamp is that certain metallic 
oxides, particularly the rare earths such as are used in the Wels- 
bach mantle, while nonconductors at ordinary temperatures, con- 
duct fairly well when hot. This conduction is of an electrolytic 
nature, so that the " glowers " endure best when used on alternating 
current of a fairly high frequency. The fundamental principle of 
the Nernst lamp is the use of a glower of such material artificially 
heated at the start to render it a conductor and then allowed to 
glow under the passage of the current. Many materials are avail- 
able for the glower, and its composition has been changed from 
time to time. In one of Nernst's original patents the composition 
was specified as of zirconia, erbia, and yttria. The later composi- 
tions have substituted thoria in part or wholly for the zirconia, and 
have also included the ceria that is found so effective in Welsbach 
mantles. Recent glowers are of a mixture of ceria, thoria and zir- 
conia. Whatever the exact composition employed, the procedure 
in manufacture is quite similar to that used in some of the later 
lamp filaments, the active material being mixed with a binder, 
formed into slender rods, and then fired at a high temperature 
until nothing but the mixed oxides are left. The glower bodies 
themselves are one-sixty-fourth to one-thirty-second of an inch in 



diameter and about one inch in effective length, the ends being 
tiny balls in which the leading-in wires are embedded. The glower 
material varies enormously in resistance with the temperature. 
Fig. '67 shows, from some of Nernst's own tests, the extent of this 
variation in specific resistance. It will be seen that at ordinary 
temperatures the glower is practically an insulator, while at a 
white heat it is a very tolerable conductor. On account of this 
very large negative temperature coefficient, the lamp would tend 
to great instability were it not for the presence of a ballast resist- 
ance in series with the glower. This ballast resistance is composed 








d ° 



ooo c 


- ~ 



1000 2000 3000 

Ohms per Cubic Centimeter 

Fig. 67. — Curve of Resistance Variation. 


of iron wire, which has a large positivd temperature coefficient, 
sealed into glass tubes filled with hydrogen to prevent oxidation. 
For starting the lamp a heating resistance close above the glower 
is provided which takes all the current when the lamp is first 
turned on and is automatically cut out of circuit when the glower 
has come to its conducting temperature. Figure 68 shows the dia- 
grammatic connections of a Nernst lamp containing three glowers. 
The apparatus consists of the glowers, the ballast tubes in series 
with them, the heater, made of wire protected by enamel, and in 
shunt with the glowers and ballast, and finally a cut-out magnet 
to remove the heater from circuit when the glowers are in action. 



The larger Nernst lamps are provided with independent terminals, 
but the single-glower lamps screw into an ordinary lamp socket 
and are treated practically like any other incandescent lamp. 
Fig. 69 shows one of these single-glower lamps complete. It con- 
sists of a small housing attached to an ordinary lamp base, which 
housing contains the cut-out magnet and ballast and carries a shade 
supporter with a 3- or 4-inch opal ball. The burner proper, Fig. 70, 
carrying the glower and heater, screws into the base of the housing, 
making automatic connection as it goes home. The mechanical 

' I| J- Lamp Terminals 

equipment is therefore reduced to the simplest possible terms, and 
the care required to replace a glower is no more than is demanded 
in screwing in an ordinary incandescent lamp. When turned on 
the glower comes to full incandescence in 15 to 30 seconds, and 
the lamp in this recent form is almost as convenient and workable 
as any other incandescent lamp. As now manufactured, Nernst 
lamps can be worked either on alternating- or direct-current cir- 
cuits. The glowers for the latter are of special ■ construction and 
must be burned at a definite polarity. They are readily adapted 
to high voltages, and all except the smallest sizes are manufactured 


Big. 69. — Singlc-glowcr Westinghouuc Nernst Lamp. 

Fig. 70. — Woatingriouse Nernst Screw Burner with Globe Removed. 



for 220 volts. The following table gives the rating and perform- 
ance of the sizes of Nernst lamp now in use from the manu- 
facturer's data: 












Actual Wattage 
under Test. 

88 " 
110 " 
132 " 

4 inches 

4 " 

5 " 

6 " 
8 " 
8 " 

8 " 












69 (110 v.) 


The effective life claimed for the glowers is 600 hours for 
the direct-current glower and 800 hours for alternating-current 
glowers at 60 cycles or above. At 25 cycles this life is reduced 
to about 400 hours. The heaters and ballast are stated to last 


about 3000 and 15,000 hours respectively, and are easily replaced 
when necessary. Fig. 71 shows the distribution of light in the 
lower hemisphere of the single-glower lamps referred to in the 
table. The temperature reached by the Nernst glower being 
upwards of 2000° C, the light is of good color, about the same 
as that from a tungsten lamp, and the efficiency is fairly high, as 
will be seen from the table already given. Reckoned on the basis 
of mean spherical candle power, the specific consumption, includ- 
ing the ballast, is somewhere between 1.5 and 2 watts per candle 

The structure of the lamp, however, requires that it should 
be compared with other illuminants rather on the basis of its 
mean lower hemispherical candle power, assuming the other lamps 
compared to be equipped with suitable reflectors. Aside froip its 
convenient use as a practical illuminant, it furnishes for experi- 
mental purposes one of the most convenient of light sources, since 
once the lamp is in action it can be run in any position, and 
gives a very steady and brilliant light, the intrinsic brilliancy of 
the glower being about 3000 candle power per square inch. In 
such use the current should be held accurately uniform by means 
of a milli-amperemeter, and the glower must be very carefully 
shielded from draughts, to which it is hypersensitive. 



The electric arc is the most intense artificial illuminant and 
the chief commercial source of very powerful light. A full 
account of it would make a treatise by itself, so that we can here 
treat only the phases of the subject which bear directly on its 
place as a practical illuminant. First observed, probably, by Volta 
himself, the arc was brought to general notice by Davy in 1808 
in the course of his experiments with the great battery of the 
Royal Institution. If one slowly breaks at any point an electric 
circuit carrying considerable current at a fair voltage, the current 
does not cease flowing when the conductor becomes discontinuous, 
but current follows across the break, with the evolution of great 
heat and a vivid light. If the separation is at the terminals of 
two carbon rods the light is enormously brilliant, and by proper 
mechanism can be maintained tolerably constant. The passage of 
the current is accompanied by the production of immense heat, 
and the tips of the carbon rods grow white hot, and serve as the 
source of light. In an ordinary arc lamp the upper carbon is the 
positive pole of the circuit, and is fed slowly downward, so as 
to keep the arc uniform as the carbon is consumed. The main 
consumption of energy appears to be at the tip of this positive 
carbon, which is by far the most brilliant part of the arc, and 
at which the carbon fairly boils away into vapor, producing a 
slight hollow in the center of the upper carbon, known as the 
" crater." 

The carbon outside the crater takes the shape of a blunt point, 
while the lower carbon is rather evenly and more sharply pointed, 
and tends, if the arc is short, to build up accretions of carbon into 
somewhat of a mushroom shape. Fig. 72 shows the shape of 
these tips much enlarged, as they would appear in looking at the 
arc through a very dark glass. Under such circumstances the light 
from the arc between the carbon points seems quite insignificant, 
and it is readily seen that the crater is by far the hottest and most 
brilliant region. In point of fact the crater may reach a temper- 



ature of probably 3500° to 4000° C, and gives about 50,000 candle 
power per square inch of surface — sometimes much more. 

It is obvious that the more energy spent in this crater the more 
heat and light will be evolved, and that the concentration of much 
energy in a small crater ought to produce a tremendously powerful 

Fig. 72. — The Electric Arc. 

arc. It is not surprising, therefore, to find that the larger the 
current crowded through a small carbon tip, — in other words, 
the higher the current density .of the arc, — the more intense the 
luminous effects and the more efficient the arc. Fig. 73 shews this 
fact graphically, giving the relation between current density and 
light in an open arc maintained at uniform current and voltage. 


The change in density of current was obtained by varying the 
diameter of the carbons employed, the smallest being about five- 
sixteenths inch in diameter, the largest three-fourths inch. The 
current was 6.29 amperes, and the voltage about 43.5. The effi- 
ciency of the arc appears from these experiments to be almost 
directly proportional to the current density. But if the carbon is 
too small it wastes away with inconvenient rapidity, while if it be 
too large the arc does not hold its place steadily and the carbon 
gets in the way of the light. 


Mean Hemispherical c.p. Lower Hemisphere 

Fig. 73. — Relation between Current Density and Intensity. 

The higher the voltage the longer arc can be successfully worked, 
but here again there are serious limitations. With too short an 
arc the carbons are in the way of the light, and the lower carbon 
tends to build up mushroom growths, which interfere with the 
formation of a proper arc. In arcs worked in the open air the arc 
is ordinarily about an eighth of an inch long. If the voltage is 
raised above the 40 to 45 volts at the arc commonly employed for 
open arcs, the crater temperature seems to fall off and the arc gets 
bluish as it lengthens from the larger proportion of light radiated 
by the glowing vapor between the carbon poles. 

So it comes about that commercial arcs worked in the open air 
generally run at from 45 to 50 volts, and from 6 to 10 amperes. 


The softer and finer the carbons the lower the voltage required to 
maintain an arc of good efficiency and proper length, so that arcs 
can be worked successfully att 25 to 35 volts with proper carbons, 
and with very high efficiency, but at the cost of burning up the 
carbons rather too rapidly. Abroad, where both high-grade car- 
bons and labor are cheaper than in this country, such low-voltage 
arcs are freely used with excellent results, and give a greatly in- 
creased efficiency. Sometimes three are burned in series across 
110-volt mains, where in American practice one, or at most two, 
arc lamps would be used, in series with a resistance coil, the same 
amount of energy being used in each case. With proper carbons 
too, a steady and efficient arc can be produced taking only 3 or 4 
amperes, and admirable little arc lamps of such kind are in use on 
the Continent. The carbons are barely as large as a lead pencil 
and the whole lamp is proportionately small, but the light is 
brilliant and uniform. 

The upper carbon burns away about twice as fast as the lower, 
and the rate of consumption is ordinarily from 1 to 2 inches per 
hour, in commercial open arc lamps. 

The carbons themselves are generally about one-half inch in 
diameter, and one or both are often cored, i.e., provided with a 
central core, perhaps one-sixteenth inch in diameter, of carbon 
considerably softer than the rest. This tends to hold the arc cen- 
trally between the carbons and also steadies it by the greater mass 
of carbon vapor provided by the softer portion. Generally it is 
found sufficient to use one cored and one solid carbon in each arc, 
although in this country arcs burning in the open air usually are 
provided with solid carbons only. 

In American practice such open arcs have almost passed out of 
use, being replaced by the so-called inclosed arcs and by the still 
more recent luminous and flaming arcs. During the past ten years 
all these have gone into use in immense numbers, until at the 
present time the open arc is very rarely installed, and illuminating 
companies are discarding them as rapidly as they find it convenient 
to purchase improved equipment. 

The principle of the inclosed arc is very simple. It merely con- 
sists in fitting around the lower carbon a thin, elongated vessel of 
refractory glass with a snugly fitting metallic cap through which 
the upper carbon is fed, the fit being as close as permits of proper 
feeding. The result is that the oxygen is quickly burned out of the 


globe, and the rapid oxidation of the carbon ceases, the heated gas 
within checking all access of fresh air save for the small amount 
that works in by diffusion through the crevices. 

The carbon wastes away at the rate of only, something like one- 
eighth inch per hour under favorable circumstances, and the lamp 
only requires trimming once in six or eight full nights of burning, 
instead of each night. For all-night lighting it used to be neces- 
sary to employ a double-carbon lamp, in which were placed two 
pairs of carbons, so that when the first pair was consumed the 
second pair would automatically go into action and finish out 
the night. The inclosed lamp burns 100 hours or more with a 
single trimming. Even much longer burning than this has been 
obtained from a 12-inch carbon, such as is customarily used, but 
one cannot safely reckon on a better performance without very 
unusual care. 

Fig. 74 shows a typical inclosed arc lamp, of the description 
often used on 110-volt circuits, both with and without its outer 
globe and case. The nature of the inner globe is at once apparent, 
but it should also be noted that the clutch by which the carbon is 
fed acts, as in many recent lamps, directly upon the carbon itself, 
thereby saving the extra length of lamp required by the use of 
a feeding rod attached to the carbon. Finally, at the top of the 
lamp is seen a coil of spirally wound resistance wire. The pur- 
pose of this is to take up the difference between 110 volts, which 
is the pressure at the mains, and that voltage which it is desired 
to use at the arc and in the lamp mechanism. 

Such a resistance evidently involves a considerable waste of 
energy, but in the inclosed arc the voltage at the arc itself is, of 
necessity, rather high, 70 to' 75 volts, so that the waste is less than 
it would otherwise be. 

It has been found that when burning in an inner globe without 
access of air, the lower or negative carbon begins to act badly, and 
to build up a mushroom tip, when the voltage falls below about 
65 volts. Hence it is necessary to the successful working of the 
scheme that the arc should be nearly twice as long as when the 
carbons are burning in open air. This has a double effect, in 
part beneficial, in part harmful. With the increased length the 
crater practically disappears, and the light is radiated very freely 
without feeing blocked by the carbons. Hence the distribution of 
light from the inclosed arc is better than from an open arc. 


On the other hand, there is no point of the carbon at anything 
like the temperature of the typical open-arc "crater," and the 
intrinsic efficiency is thereby greatly lowered. Also, if the inclosed 
arc is to take the same energy as a given open arc, the current in 
the former must be reduced in proportion to the increased voltage, 
hence, other things being equal, the current density is lowered, 
which also lowers the efficiency. 



Fig. 74. — Typical Inclosed Arc Lamp. 

The compensation is found in the lessened care and the lessened 
annual cost for carbons. The carbons themselves have to be of 
a special grade, and are about two and a half times as expensive 
as plain solid carbons, but the number used is so small that the 
total cost is low. There is some extra expense on account of 
breakage of the inner globes, but the saving in labor and carbons 
far outweighs this. Moreover, the light, while decidedly bluish 
white, is much steadier than that of the ordinary open arc, and the 
inner globe has material value in diffusing the light, being very 



often of opal glass, so that the general effect is much less dazzling 
than that of an open arc, and the light is better distributed. 

In outdoor lighting the greater proportion of nearly horizontal 
rays from the inclosed arc is of considerable benefit, while in build- 
ings the same property increases the useful diffusion of light, as will 
be presently shown. Of course, when inclosed arcs are operated in 
series, as in street lighting, the resistance of Fig. 74 is reduced to 
a trivial amount, or abolished, so that the extra voltage required 
with the inclosed arc is the only thing to be considered. The in- 
closed arc used in this way is very materially steadier as an illu- 
minant than an open arc taking the same current, and experience 
shows that it may be substituted for an open arc, taking about the 
same energy, with general improvement to the illumination. 

The weak point of the open arc is its bad distribution of light, 
which hinders its proper utilization. The fact that most of the 
light is delivered from the crater in the upper carbon tends to 
throw the light downward rather than outward, and much of it is 
intercepted by the lower carbon. Fig. 75 gives from Wybauw's 

experiments the average dis- 
tribution of light from 26 dif- 
ferent arc lamps, representing 
the principal American and 
European manufacturers. 
The radii of the curve give 
the intensities of the light in 
various angles in a vertical 
plane. The distribution of 
light in space would be nearly 
represented by revolving this 
curve about a vertical axis 
passing through its origin, 
although at any particular 
moment the distribution of 
light from an arc may be far from equal on the two sides. 

The shape of the curve is approximately a long ellipse with its 
major axis inclined 40 degrees below the horizontal. The presence 
of globes on the lamps may modify this curve somewhat, but in 
ordinary open arcs it always preserves the general form shown. 
The small portion of the curve above the horizontal plane shows 
the light derived from the lower carbon and the arc itself, while 

80° 00" 

Fig. 75. — Distribution of Light from 
an Open Arc. 



the major axis of the curve measures the light derived from the 
crater. The tendency, then, of the open arc is to throw a ring of 
brilliant light downward at an angle of 40 degrees below the hori- 
zontal, so that within that ring the light is comparatively weak, 
and without it there is also considerable deficiency. Hence the 
open arc, if used out of doors, fails to throw a strong light out along 
the street, while the illumination is strong in a zone near the lamp. 

For the same reason the open arc is at a disadvantage in interior 
lighting, for the reason that most of the light, being thrown down- 
ward, falls upon things and surfaces far less effective for diffusion 
than the ordinary walls and ceiling. Hence one of the very best 
ways of using arcs for interior lighting is to make the lower carbon 
positive instead of the upper, and to cut off all the downward light by 
a reflector placed under the lamp. Then practically all the light is 
sent upward and outward to be diffused by the walls and ceiling. 

The inclosed arc, on the other hand, gives a much rounder, fuller 
curve of distribution, the light being thrown well out toward the 
horizontal, and there is also a 
pretty strong illumination above 
the horizontal. 

Fig. 76 shows a composite dis- 
tribution curve from ten or a 
dozen inclosed arc lamps, such as 
are used on constant-potential 
circuits, including various makes. 
Most of them were lamps taking 
about 5 amperes, and therefore 
using nearly 400 watts at the arc, 
besides the energy taken up in the 
resistance and the mechanism. 
These figures include the inner 
globe, of course, generally of opal 
glass, which is of some benefit in 
correcting the strong bluish tinge which is produced by the long arc. 
After a few hours' burning a slight film collects on the inner globe, 
which tends to the same result. 

As ordinarily employed, inclosed arc lamps take from 5 to 7 
amperes, although now and then 3- or 4-ampere lamps are used. 
These smaller sizes are very unsatisfactory in the matter of color of 
the light, and are not widely used. 

80° 50° 



J *s s^*^ J*^\ \ ^JL- \ 1 1 1 m 1 

r+gL S j-^\^ 1 f^ \ \^~— k 1 1 1 1 1 & 1 

*&?ii' ,'^[ i \ l 1 1 ill IV 1 

*'T-' 1 111 11 1 111 ^ 1 



H 160 I 



20 ( 

10 s 

10 c 

20 c 


80° 70 c 00 c 50° 40' 

Fig. 76. — Distribution of Light 
from Inclosed Arc. 


Outside of America the inclosed arc is little used, for abroad labor 
is much cheaper than here, and carbons of a grade costly or quite 
unattainable here are there reasonably .cheap, so that the consider- 
ably higher efficiency of the open arc compensates for the extra 
labor and carbons. Aside from this, the bluish tinge of the light 
from inclosed arcs of small amperage is considered objectionable, 
and the gain in steadiness so conspicuous in American practice 
almost or quite disappears when the comparison is made with open 
arcs taking the carbons available abroad. 

At its best the carbon electric arc has fully three times the 
efficiency of a first-class carbon incandescent lamp, but this ad- 
vantage is somewhat reduced by the need of diffusing globes to 
keep down the dazzling effect of the arc and to correct the distri- 
bution of the light. Taking these into account, and also reckoning 
the energy wasted in the resistances in case of arc lamps worked 
from constant-potential circuits, the gain in efficiency is con- 
siderably reduced, and if one also figures the better illumination 
obtained by using distributed lights in incandescent lighting, 
the arc lamp has a smaller advantage than is generally supposed. 
Many experiments bearing on this matter have been made, and 
a study of the results is highly instructive. 

By far the most complete investigation of the properties of 
the inclosed type of arc lamps is that made by a committee of the 
National Electric Light Association a few years ago. The investi- 
gation was upon the arc lamps both for direct and alternat- 
ing currents, as customarily used on constant-potential circuits. 
The results, however, are not materially different, so far as dis- 
tribution of light goes, from those that belong to similar lamps 
for series circuits. Fig. 76 is the composite curve of distribu- 
tion obtained by this committee in the tests of direct-current 

The weak point of such lamps as efficient illuminants lies in the 
large amount of energy wasted in the lamp mechanism, including 
the resistance for reducing the voltage of the mains to that desir- 
able for the inclosed arc. This loss amounts ordinarily to nearly 
30 per cent of the total energy supplied, so that while the arc itself 
is highly efficient the lamps as used are wasteful. No one but 
an American would think of working a 75-volt arc off a 120-volt 
circuit and absorbing the difference in an energy-wasting resist- 
ance, but the advantages of the inclosed arc are so great in point 


of steadiness and moderate cost of labor that the bad practice 
has been considered commercially advantageous. 

At present alternating-current arc lights are being rather widely 
used, both on constant-potential and on constant-current circuits, 
and such arcs present some very interesting characteristics. Evi- 
dently when an arc is formed with an alternating current there 
is no " positive 7 ' and no "negative" carbon, each carbon being 
positive and negative alternately, and changing from one to the 
other about 7200 times per minute — 120 times per second. 

Under these circumstances no marked crater is formed on either 
carbon, and the two carbons are consumed at about an equal rate. 
As a natural result of the intermittent supply of energy and 
the lack of a localized crater, the average carbon temperature is 
considerably lower than in case of the direct-current arc, and 
the real efficiency of the arc as an illuminant is also much lowered. 
Tests made to determine this difference of efficiency have given 
somewhat varied results, but it seems probable that for unit energy 
actually applied to the arc itself the direct-current arc will 
give somewhere about 25 per cent more light than the alternating- 
current arc. But since when working the latter on a constant- 
potential circuit the surplus voltage can be taken up in a reactive 
coil, which wastep very little energy, instead of by a dead resist- 
ance, which wastes much, the two classes of arcs then stand upon 
a more even footing than these figures indicate. This comparison 
assumes inclosed arcs in each case. 

The chief objection to the alternating-current arc has been the 
singing noise produced by it. This is partly due to the vibration 
produced in the lamp mechanism and partly to the pulsations 
impressed directly on the air by the oscillatory action in the arc 
itself. The former can be in great measure checked by proper 
design and manufacture, but the noise due directly to the arc is 
much more difficult to suppress. 

Abroad, where, for the reason already adduced, open arcs are 
commonly used, a specially fine, soft carbon is used for the 
alternating arcs, and the noise is hardly perceptible. These soft, 
volatile carbons, particularly when used at a considerable current 
density, give such a mass of vapor in the arc as to endow it with 
added stability and to muffle the vibration to a very marked 
degree. The result is a quiet, steady, brilliant arc of most excellent 
illuminating power. But in this country such carbons are with 



difficulty obtainable, and, even if they were to be had at a reason- 
able price, could not be used in inclosed-arc lamps on account of 
rapid smutting of the inner globe. 

In selecting alternating-current lamps for indoor work, great 
care should be exercised to get a quiet lamp. Some of the Ameri- 
can lamps when fitted with tight outer globes and worked with a 
rather large current are entirely unobjectionable, but in many 

90° 80° 70° 60° 60' 40° 

Fig. 77. — Distribution from Alternating Inclosed Arc. 

cases there is noise in the mechanism, or the globe serves as a 
resonator. With a current of 7 to 7.5 amperes, and a well-fitted 
and nonresonant globe, little trouble is likely to be experienced. 
Out of doors, of course, a little noise does not matter. 

The chief characteristic of the alternating arc, as regards dis- 
tribution of light, is its tendency to throw its light outward rather 
than downward like the direct-current arc; in fact, considerable 
light is thrown above the horizontal, which materially aids 


For this reason it is often advantageous to use reflecting shades 
for such lamps, so as to throw the light out nearly horizontally 
when exterior lighting is being done. Indoors, diffusion answers 
the same purpose, unless powerful downward light is needed, when 
the reflector is of service. 

Fig. 77, from the committee report already mentioned, shows 
the distribution of light from an alternating-current lamp fitted 
with a porcelain reflecting shade, with an opalescent outer globe, 
and with a clear outer globe.' The abolition of the outer globe 
and the use of the reflector produce a prodigious effect in 
strengthening the illumination in the lower hemisphere, and this 
hemispherical illumination is for some purposes a convenient way 
of reckoning the illumination of the lamp. But a truer test is 
the spherical candle power, since that takes account of all the 
light delivered by the lamp. Alternating arc lamps seem to work 
best at a frequency of 50 to 60 cycles per second. Above 60 
cycles they are apt to become noisy, and below about 40 cycles 
the light flickers to a troublesome extent. The light of the alter- 
nating arc is really of a pulsatory character, owing to the alterna- 
tions. A pencil rapidly moved to and fro in the light of such an 
arc shows a number of images — one for each pulsation; and this 
effect would be very distressing if one had to view moving objects, 
like quick running machinery, by such light. A harrowing tale is 
told of a certain theater in which alternating arcs were installed 
for some gorgeous spectacular effects, and of the extraordinary 
centipedal results when the ballet came on. 

This pulsation is somewhat masked when the inclosed arc is 
used, even with a clear outer globe, and is generally rather incon- 
spicuous when an opal outer globe is used. It is also reduced 
when a fairly heavy current (7 to 8 amperes) is used, and when 
very soft carbons are employed, as they can be in open arcs. 

An interesting comparison of direct-current and alternating- 
current inclosed arcs, as used on constant-potential circuits, is 
found in the following table, from the report already quoted. 

It must be remembered that the results are in Hefner units. 
This unit is exactly 0.9 candle power, so that the mean results, 
reduced to a candle-power basis, are, for efficiency when using 
clear outer globes, as follows: 

Direct-current arc 2.89 watts per candle power. 

Alternating arc 2.96 watts per candle power. 



— 1 

D. C. Lamp. 


Watts Consumed. 

Mean Intensity in 
Heat Units. 

Mean Watts. 





Heat Unit*. 










2£ i 

6 8 







362 - 















. 195 



















































































A. C. 


















































6 20 























2 19f 










































* Condition of no outer globe, t Condition with shade on lamp. 
Note. — All marked values not included in the mean. 

These efficiencies are on their face but little better than those 
obtained from incandescent lamps. There is little doubt that as 
a matter of fact a given amount of energy applied even to 3-watt 
incandescent lamps will give more useful illumination than if used 
in arcs of the types here shown. The incandescents lose some- 
what in efficiency, but gain by the fact of their distribution in 
smaller units. In comparison with tungsten lamps these arcs are 
hopelessly outclassed and are now obsolescent. 

But for some purposes the arcs are even now preferable on 
account of their whiter light and the very brilliant illumination 
that is obtainable near them. 

Both direct- and alternating-current inclosed arcs gain, by the 
use of rather large currents, both in steadiness and in efficiency, 
and moreover give a whiter light. The same is true, for that 
matter, of open arcs, in which the larger the current the higher 


the efficiency. Very many experiments on the efficiency of open 
arcs have been made, with moderately concordant results. Their 
efficiency ranges in direct-current arcs from about 1.25 watts per 
candle in the smallest to about 0.6 or a little less in the most 
powerful. Fig. 78 shows considerable number of results by dif- 
ferent experimenters consolidated into a curve giving the relation 
between current and efficiency, as based on mean spherical candle 

There is generally accounted to be about 25 per cent difference 
in absolute efficiency in favor of the continuous-current arc. 

Fig. 78. — Relation between Current and Efficiency. 

Within the past three or four years a good many so-called 
intensive arcs have come into use, at first abroad and later in 
this country. These are practically inclosed area worked at very 
high current density. They are usually small lamps for indoor 
use taking from four to five amperes and having electrodes one- 
fifth to one-fourth of an inch in diameter instead of the usual 
one-half inch or thereabouts. 

The result is, as would be anticipated from Fig. 78, a very great 
increase in efficiency, since the current density is about four or 
five times greater than usual in the earlier arc lamps. There ia 


an additional advantage gained from the small carbons in that 
the crater is less marked and less sheltered, almost the entire 
end of the carbon being raised to a very vivid incandescence. 
These arcs are usually fitted with a single small opal globe, al- 
though sometimes a clear inner globe and an exterior opal globe 
are employed. 

The specific consumption of such arcs commonly runs from 1 
to 1.5 watts per mean lower hemispherical candle power, which 
places them in the same class with regard to efficiency as the 
recent metallic filament incandescent lamps. The intensive arc, 
however, has an enormous advantage in the matter of color. Even 
the most efficient of the metallic filament incandescents are still 
considerably off white, while the intensive arc is a sufficiently 
close approximation to sunlight for almost any practical purpose, 
and is by far whiter than any other illuminant suitable for general 
commercial purposes. For use in shops where color matching has 
to be done, the intensive arc is the only efficient illuminant as 
yet available which gives approximately sunlight values to the 
colors, and by reason of this advantage it has come into extended 
use and is rapidly driving out the older illuminants in cases where 
critical color matching is important. The intensive arcs require 
a little more care than their predecessors, inasmuch as the mech- 
anism is somewhat more delicate, and the life of the carbons is 
from 20 to 50 or 60 hours instead of at least double this period as 
in the ordinary inclosed arcs. The difference in steadiness, color, 
and efficiency, however, is so great as to leave no comparison 
between the two for practical indoor illumination. 

With the exception of the intensive arc just mentioned, all forms 
of carbon arc are rapidly becoming obsolescent in this country. 
They have indeed little excuse for existence from the standpoint 
of efficiency since the introduction of arcs of the so-called luminous 
and flaming types. These are widely separated from previous arcs 
in that their light is due, not mainly to the high incandescence of 
the electrodes, but to the incandescent vapors of the arc stream 

The flaming arcs employ electrodes charged with easily vaporiz- 
able metallic salts which give intense light from the vapor streaming 
between the poles. This light, like other light from incandescent 
vapors, shows a discontinuous spectrum. The efficiency of the 
light emitted, therefore, does not depend, as in the case of incandes- 


cent solids, upon the absolute temperature, but rather upon the 
special character of the vapor; and as a result it is thus feasible to 
obtain luminous efficiencies very much higher than could practically 
be reached by any incandescent solid. The substances used for 
mineralizing the electrodes are very various. The chief one is cal- 
cium fluoride, which possesses the somewhat unusual property of 
giving a discontinuous spectrum as a compound. It is sometimes 
mixed with the analogous strontium and barium fluorides, in order 
to modify the color, strontium giving a ruddy tinge and the addi- 
tion of barium making the light somewhat whiter. The calcium 
fluoride by itself gives a brilliant golden-yellow light of very high 
value in luminosity, but yet of too strong hue to be altogether 
pleasing, so that the lamp using such electrodes is better suited 
for outdoor work than for interior work. 

In fact, all the flaming and luminous arcs give off a considerable 
amount of solid fumes composed of oxides of the metals concerned, 
which are somewhat objectionable in interior lighting. Carbon 
possesses for such work the unique advantage of producing an 
oxide that is gaseous, colorless, and odorless. 

Some electrodes for flaming arcs are also charged with a certain 
quantity of rare earths, by-products of Welsbach mantle manu- 
facture, which give a nearly white flame. All the substances here 
mentioned have white or yellowish-white oxides, a point of some 
practical importance with respect to the results of the fumes 

Still another group of flaming arcs, usually known in this country 
as luminous arcs, are charged with compounds of iron, titanium, 
and chromium in various proportions, some minor constituents 
being occasionally added to these. These arcs, the spectra of the 
metals being enormously rich in widely distributed lines, give a 
much whiter light than the arcs charged with the calcium group 
of metals, and the electrodes are consumed much less rapidly, an 
advantage which is punished by a somewhat lower efficiency. 
They also give off a smudge of dark-colored oxides difficult to take 
care of in the lamp, and require very active ventilation to keep the 
globes clear. The two groups of arcs just mentioned are of quite 
different character in some important respects and perform very 
differently in practice, though both are widely and successfully used. 

Taking up first the flame lamps proper, in which the electrodes 
are carbon sticks, one or both of them "mineralized" usually with 


calcium fluoride, the early types of these lamps were commonly 
arranged for the use of inclined electrodes forming an acute angle 
with each other and fed down through the lamp casing into a globe 
at the extreme bottom of the lamp. Fig. 79 gives the appearance 
of a typical lamp of this kind and Fig. 80 shows the arrangement 

Fig. 79. Hg. 80. 

of the electrodes and mechanism. The two carbons, converging, 
meet in a cup-ehaped hollow at the extreme bottom of the lamp 
casing lined with refractory material. The arc is struck by swing- 
ing one of the carbons slightly and is kept in place within the cup 
and prevented from running up on the carbons by the repulsion 
of a slight magnetic field established by a magnet near the bottom 


of the casing. This is the same principle employed in the old 
Jamin candle in use more than thirty years ago. 

Such flame arcs are made both for direct and alternating current. 
In the former case sometimes only one carbon, the positive one, is 
mineralized, the negative electrode being a plain carbon stick of 
somewhat smaller diameter, so that the electrodes will burn away 
uniformly. Sometimes the mineralized carbon is furnished with 
a slender wire as a core to increase the conductivity. In some 
cases both electrodes have been mineralized, the amount and dis- 
tribution of the mineral material having varied greatly in different 
lamps. The electrodes are long, commonly from 18 to 24 inches, 
and burn at the rate of 1 to 1J inch per hour, so that the burning 
life is commonly from 14 to 18 hours. The voltage at the arc is 
40 to 45, and the current is usually 10 to 12 amperes. 

Such lamps are adapted to run two in series on ordinary multiple 
circuits. Each lamp gives an output of approximately 1200 to 
1400 mean spherical candle power at a specific consumption of 0.4 
to 0.5 watt per mean spherical candle power, reckoning the energy 
at the lamp terminals so to include the small steadying resistance 
which is needful in case of multiple lamps worked in this way. 
These figures are true only for the lamp charged with calcium 
fluoride, those modified to give a ruddy or whiter light having 
materially less efficiency. The alternating-current lamps of similar 
type operate at a specific consumption of between 0.6 and 0.7 
watt per mean spherical candle power. All these figures apply 
to the lamp, as usually equipped, with an opal globe. 

These inclined carbon flame arcs have come into very wide use 
in this country mainly for display lighting and the illumination of 
very large interiors; abroad, for both commercial and street light- 
ing. Fig. 81 shows a typical distribution curve from a direct- 
current lamp of this class. 

More recently another type of flame arc, primarily due to Pro- 
fessor Blondel, has been very successfully introduced. In this the 
carbons are vertical, as in an ordinary arc lamp, the heavily 
mineralized positive carbon being below, while usually a plain 
carbon is employed ajx>ve. The positive carbon in this, as in the 
previous lamps, is the larger in diameter, and the lamp is furnished 
with a focusing feed, so that the arc is maintained in one position 
just below a little cup of refractory material through which the up- 
per carbon passes. Lamps of this class have shown most extraor- 



dinary results in efficiency, and the vertical carbons give a better 
curve of distribution for outside work than do the converging 
carbons. The voltage at the arc is ordinarily a scant 40 volts, and 

the lamps are burned two in series on multiple circuit. They can 
also, like other flame lamps, be conveniently adapted for use on 
scries circuits if desirable. They work 
well on a somewhat smaller current than 
is usual with the inclined carbon lamps. 
Fig. 82 shows the Blondel lamp as 
manufactured by Crompton & Co. This 
is made both with single carbons and 
with double carbons for longer burning, 
following the practice of the earlier open 
arcs. The consumption of the electrodes 
is less rapid than in the usual electrodes 
for inclined lamps, amounting to 0.8 inch 
per hour or less according to current. 
One pair of 15-inch carbons lasts about 
17 hours at 10 amperes and about 22 at 
7 amperes. The electrodes generally 
used give the same yellowish light as 
other ordinary flame- arc electrodes. 
Fig. 83 shows the actual results of tests 
of a lamp of this type arranged for series 
burning at a current of 6.6 amperes. The watts per mean spherical 
candle power were 0.46, while the watts per mean lower hemi- 

Fig. 82. 



spherical candle power dropped to 0.326, a very extraordinary per- 
formance for an arc taking only about 260 watts at the terminals 
and equipped with an opal globe. Similar arcs at 8 to 10 amperes 
give an efficiency materially higher even than this. 

The chief practical trouble with the name are being the necessity 
of frequent trimming, the attention of inventors haa been lately 
drawn toward the production of long-burning electrodeB, either by 
increasing the length or cross section or slightly modifying the 

Fig. 83. 

composition. The first and perhaps best known of the long-burning 
lamps is the Jandus regenerative flame lamp of which the general 
appearance is shown in Fig. 84 and the cross section in Fig. 85. 
The peculiarity of the lamp is the provision of two cooling cham- 
bers, of cast iron enameled white, which pass outside the globe 
and connect with the lamp casing above and below. The arc itself 
burns in an inner clear-glass flue surrounded by the ordinary opal 
globe. Air is admitted from the bottom and the fumes from the 
lamp pass upward and outward into the cooling flues, where they 


are deposited, so that the chimney and globe are kept reasonably 
free from them, and consequently the arc can burn for a much 
longer period without producing an 
opaque coating. 

The lower, heavily mineralized, 
electrode is a carbon stick of stell- 
ate cross section. The active ma- 
terial is packed into the eight 
channels between the eight arms of 
the star, the electrode being about 
one inch in diameter. The upper 
negative electrode is a cored carbon 
stick. These lamps are adapted for 
currents of 5 to 7 amperes, and, 
owing to the comparatively com- 
plete inclosure of the arc, the arc 
voltage runs high, — 70 to 90 volts, 
— so that the lamp with a small 
steadying resistance can be burned 
singly on constant-potential cir- 
cuits if it is desirable. It also lends 
itself readily to use on series circuits. Fig. 86 shows the distri- 
bution curve of such a lamp worked on a series circuit of 6.7 
amperes. It will be seen that the distribution is a favorable one 
for outside lighting, and the efficiency is high, the specific con- 
sumption reckoned at the terminals of the lamp being 0.58 watt 
per mean spherical candle power and 0.36 watt per mean lower 
hemispherical candle power. The life of the electrodes is approxi- 
mately 75 hours at this current, and it should be noted that 
the condensing chambers, owing to the intensity of the surface 
radiation from the opal globe, do not cast a noticeable shadow. 

An interesting type of Same-arc lamp for series circuits has 
recently been introduced by the General Electric Company in 
this country and is employed to some extent in street illumina- 
tion, particularly in the case of open squares. This is a vertical 
carbon lamp with electrodes of a slightly different composition from 
those heretofore mentioned, and especially adapted to work on 
moderate currents. Fig. 87 shows the distribution curve of this 
light as used in street lighting practice in Boston, Mass., equipped 
with a fairly dense Alba globe and enameled reflector. It is inter- 


esting, from the exceptionally good distribution, for lighting large 
areas. The specific consumption with this particular globe is 0.35 
watt per mean lower hemispherical power. This represents the 
ordinary burning condition of the lamp, no pains being taken to 
keep the globe free from the deposit. With a clear or very light 
opal globe kept rigorously clean during a test the efficiency figures 
would run somewhat higher. The life of the electrodes is about 
20 hours. 

Fig. 86. 

Within the past two or three years some remarkable white-flame 
electrodes, under the name of Alba, or T. B., have been produced 
by Siemens and Halske, to which reference has already been made. 
These have very nearly as high an efficiency as the best of the 
yellow-flame electrodes, and are adapted for burning in various 
forms of flame arcs. They are widely employed abroad for street 



lighting in the Siemens and Halske vertical carbon lamps, and 
give at very high efficiency a light that in color is quite indistin- 
guishable from that of a first-class carbon arc. All the flame lamps 
suffer in this country from a strong prejudice against frequent 
trimming and from the high cost of the electrodes, which are not 
yet produced in large amounts here and are heavily punished by 
the customs duties. 

To meet American requirements, a radically different type of 
arc, commonly known as the luminous arc, has come into very 
extensive use. This is essentially a flame arc, but the active 

77v\A\ \ VJ 

Fig. 86. 

material is, as already indicated, of very different character. The 
best-known form of luminous arc is the so-called magnetite arc, 
in which the lower (negative) electrode consists of an iron tube 
packed with a mixture of magnetite, titanium oxide, and some- 
times small quantities of chromium oxide. Approximately 75 per 
cent of the mixture is magnetite and nearly all the rest titanium 
oxide. Magnetite is a pretty fair conductor, vaporizes easily, 
giving a good volume of vapor in the arc, and while by itself it is 
not a very efficient illuminant, it serves as an effective carrier for 
the titanium, to which much of the brilliancy of the arc is due. 



Titanium oxide by itself is a bad conductor, vaporizes with consid- 
erable difficulty, and slags abominably, so that it is impracticable 
to use a large percentage in connection with the magnetite. 

The mixture is quite sensitive, in light-giving properties and 
steady-burning quality, to small changes of composition. The posi- 
tive electrode, commonly the upper one, although in lamps of some 
makers the position is reversed, is a short copper cylinder which 
burns away very slowly and does not visibly color the light. 
The arc stream is most intense near the surface of the negative 
electrode, and the light falls away considerably toward the 

Neither electrode gives any material amount of light by incan- 
descence. The shape of the arc stream causes the magnetite lamp 
to give an exceptionally large proportion of its light at or near 
the horizontal, and it is usually worked with a reflector to turn 
downward some of the beams which would naturally pass above 
the horizontal. It has a distribution, therefore, most convenient 
for street lighting, for which it has come into very great use. The 
magnetite lamp is made for the most part in two sizes, one taking 
about 4 amperes and the other about 6.6 amperes; and the lamps 
arc usually worked in series, either from arc-light generators or 
more commonly from mercury rectifiers. 



Fig. 88 shows the distribution curve of the ordinary 4-ampere 
magnetite lamp furnished with the commonly employed five- 
eighths-inch magnetite lower electrode. This lamp took about 
310 watts and gave 467 mean lower hemispherical candle power 
and 237 mean spherical candle power. The specific consumption, 
therefore, was 0.66 watt per candle for the former case and 1.31 
watts per candle for the latter. This performance was with a 
clear globe frosted on the bottom and fitted with the usual ash 
pan below the lower electrode. The life of the electrodes in such 
a lamp is from 150 to 200 hours. The performance shown in 
this curve is a thoroughly typical one of average performance. 
The shape of the curve is a good one for street lighting, but arcs 
of this character give so tremendous a glare that it is better to 

Fig. 88. 

sacrifice something of the light and use opal globes with them 
when employed in a thickly settled district or where the traffic 
under them is considerable. 

Fig. 89 shows the corresponding performance of a series mag- 
netite lamp worked at 6.6 amperes. This lamp was equipped 
with a clear globe and the lower electrode was a half-inch stick 
giving a burning life of about 60 hours. The electrode was a 
particularly good one from the standpoint of efficiency, as the 
result shows. It should be clearly understood that in the case 
of luminous electrodes of all kinds much depends on the rate of 
combustion of the light-giving material. If the electrode compo- 
sition is planned to give a long life, it will, other things being 
equal, give a lower efficiency, and vice versa. This 6.6-ampere 
lamp took 510 watts while giving 1472 mean lower hemispherical 



candle power and 809 mean spherical candle power. The specific 
consumption in the former case was 0.35 watt per candle and in 
the latter 0.63. 



^JT^^Ca — \ \\ \ 




^ l0W "\^VS<^S^2^ r // "^\y 



s\. ' /C /^*«sV / 


N \ 

^w J ^^^ / ^^^+^i » f 





40 c 

Fig. 89. 

On this very powerful lamp a diffusing globe is even more 
necessary than in the case of the smaller lamp. Fig. 90 shows 
a distribution curve of the same lamp in a fairly light opal globe. 

Fig. 90. 

As will be seen at a glance, the presence of the globe cuts down 
the light very considerably, the mean spherical candle power being 
reduced 23 per cent. The curve is rounded by the diffusion and 
the maximum drops a little further below the horizontal. The 


distribution is still excellent, however, and for first-class street 
lighting the lamp. with the opal globe is much preferable to that 
with the clear globe. The watts taken were 510 as before, but 
the mean lower hemispherical candle power was reduced to 968 
and the mean spherical candle power to 622. The specific con- 
sumptions were respectively 0.53 and 0.82 watt per candle. 

The chief difficulty with magnetite lamps is the production of a 
quantity of brown oxides which have to be disposed of to keep 
them from settling on the globe and clogging the mechanism. 
This is done by a central draft tube through the lamp, which 
under ordinary circumstances carries out the fumes pretty suc- 
cessfully. Sometimes in hot and damp weather they give trouble 
by sticking to the upper electrode, causing the arc to burn un- 
steadily, and by depositing a brown smudge over the inside of the 
globe. These lamps, therefore, require some extra care in keeping 
the globes clean, but in spite of such drawbacks the magnetite 
arc is at the present time the best powerful illuminant available 
for outdoor use in this country. 

The earlier lamps gave a good deal of trouble both with the 
mechanism and by formation of slag and welding of the electrodes, 
but these difficulties have gradually been overcome, until at the 
present time the trouble from lamps being out is no more than 
it was in the case of the earlier carbon lamps. 

From the standpoint of economy and efficiency, the carbon arc 
has little reason for use as compared with the 6.6-ampere mag- 
netite arc. The 4-ampere magnetite arc is considerably less effi- 
cient than the larger size, but still admirably suited for exterior 
lighting. Unfortunately, the magnetite arc cannot be used on 
alternating current, and if only alternating current is available it 
must be rectified before use. 

A very interesting attempt to get around this difficulty is fur- 
nished by the titanium-carbide arc, which at one time promised 
to come into considerable use. This lamp was adapted for alter- 
nating current only, carried a carbon upper electrode about one 
inch in diameter, and had as lower electrode an iron tube which 
was packed with the titanium-carbide mixture. This lamp was 
especially adapted for use on low currents, the ordinary type taking 
only 2.5 amperes and 180 watts. So admirable was the titanium 
mixture in light-giving power that the efficiency even of this small 
arc ran very high, the specific consumption per mean lower hemi- 


spherical candle power being between 0.4 and 0.5 watt, and the 
distribution resembled very much that of the magnetite arc. 
Unfortunately, the titanium electrode produced a frightful smudge 
of brown oxide when the current was pushed materially above 
the figure just stated, so that ventilation became a very serious 
matter; and when the lamp was kept to 2.5 amperes or there- 
abouts it became hypersensitive to small variations in current and 
irregularities of voltage, so that at the present time this interest- 
ing and rather promising illuminant is making very little head- 
way. It is greatly to be hoped that the difficulties met in its 
development may be overcome, because so efficient a lamp em- 
ploying alternating current would be of very great value to the 
art if reduced to a thoroughly practical form. 

We now come to a totally different class of illuminants, more 
akin to arcs than to incandescents in their physical properties, 
and hence classified with arcs, but yet radically different from arcs 
of ordinary type, in that the arc stream is produced in sealed tubes 
and the light is given by relatively long columns of vapor or gas, 
not subject to oxidation, and at comparatively moderate tempera- 
ture. The best known of these illuminants is the Cooper-Hewitt 
mercury-vapor lamp, which has now come into extended use for 
the illumination of large areas. In this lamp a long glass tube 
containing a small amount of mercury is supplied with current 
through platinum leading-in wires, one of which dips in the mer- 
cury and the other of which is attached to an iron electrode. The 
mercury is the negative electrode and when the arc is once 
started, as it can be very conveniently by momentarily tilting 
the tube so that the mercury runs into contact with the iron 
electrode and then withdraws, it fills the whole interior of the 
tube with glowing mercury vapor and gives out a very brilliant 
and steady light. 

Fig. 91 shows one of the commonest forms of Cooper-Hewitt 
lamp with its complete mounting. The upper part of the lamp 
contains inductance coils, an adjustable resistance, an automatic 
tilting magnet, and, in case of lamps operating in series, a shunt 
resistance and a cut-out. The lamp shown takes 192 watts at the 
terminals, the normal current being 3.5 amperes. The light-giving 
tube is about 22 inches in length between the bulbs and of about 
1 inch bore. The mercury electrode is contained in the blackened 
bulb at the right of the illustration, the blackening being for the 


purpose of preventing the boiling and bubbling of tbe mercury in 
the bulb from making itself visible as a flicker. 

This particular size of lamp is intended to be operated two in 
scries on the '110- volt circuit. For use singly on 110 volts, tubes 
of slightly more than double the length are used. When the current 
is thrown on the lamp is tilted, and, dropping back, starts the arc, 
which, at first curiously bluish, comes in a minute or so to intense 
brilliancy and acquires a greenish cast. The light is that of the 

mercury spectrum, which for light-giving purposes consists of three 
intense lines in the yellowish and green, reenforced by a vivid blue- 
violet line which is momentarily predominant when the arc first 
starts. Red is practically alisent from the spectrum, the only red 
lines being too faint to produce any noticeable effect. 

Though the resulting color of the light is somewhat ghastly and 
plays curious tricks with colors containing red, the lamp is very pow- 
erful and steady, of moderate intrinsic brilliancy, — some 10 to 11 
candle power per square inch, — and from its being approximately 



monochromatic is particularly effective for seeing details in black 
and white. Visual acuity for ordinary reading and writing pur- 
poses is considerably enhanced under the mercury light, so that its 
usefulness for such seeing is materially greater than its nominal 
candle power would indicate. Its efficiency, however, is high, the 
tube of Fig. 91 being rated at about 300 mean lower hemispherical 
candle power including the reflector, which implies a specific con- 
sumption of 0.6 to 0.7 watt per candle. The Cooper-Hewitt lamp 
is also available for alternating current, for use with which a small 
individual rectifier is added to the auxiliary apparatus. The tube 
has a life running to many hundred hours under favorable cir- 
cumstances and if not pushed above its rated current. 

An extremely interesting and very recent development is the 
use in connection with this lamp of a fluorescent reflecting screen 
devised by Dr. Hewitt, which adds to the light the red rays absent 
from the original mercury spectrum. In other words, the fluores- 
cent reflector transforms part of the incident light into red and 
orange light, and with a sufficient area of reflector the result is a 
pretty good white. A similar result has been reached by using 
in connection with the mercury tube a certain proportion of ordi- 
nary incandescent lamps to supply the red rays, but the fluorescent 
screen is an equally efficient and much more elegant solution of 
the difficulty. 

A still more interesting and valu- 
able illuminant is the quartz-mer- 
cury lamp, which is essentially the 
same thing as the mercury-vapor 
lamp just described, except that it 
is worked intensively in a tube of 
fused quartz, which is sufficiently 
refractory to be safely worked at 
high current density and greatly en- 
hanced temperature. Such lamps 
have come into considerable use 
abroad and promise some very 
striking developments. 

Pig. 92 shows in diagram the 
arrangement of a quartz lamp as Fig - 92, 

manufactured by the Westinghouse-Cooper-Hewitt Company in 
England. The quartz tube is carried in a clear-glass globe sur- 



mounted by a small housing containing the ballast resistance, 
tilting magnet, and cut-out, the operation of the lamp being prac- 
tically as already described for the ordinary mercury-vapor lamp. 
The tube is of clear fused quartz, consisting of a terminal bulb of 
mercury at each end, connected by a vapor tube three or four inches 
long, in which the light is produced. 

Such lamps are adapted to work on either 110- or 220-volt cir- 
cuits taking about 3.5 to 4 amperes. They give a very intense 
bluish-white light which, unlike that of the ordinary mercury- vapor 
tube, contains a perceptible amount of red radiation, although not 
enough to give reds their full value when viewed under it. The 
light is very steady and the tube holds up for a very long life stated 
at something like 2000 hours on the average. The efficiency is 

Fig. 93. 

very high, but varies somewhat from tube to tube, depending 
largely on the current density to which the lamp is forced. 

Fig. 93 shows the result of a test, made by the author, of a 
quartz-mercury lamp from one of the Continental makers, in which 
the volts at the terminal of the lamp were 224, the amperes 3.5, 
and the mean lower hemispherical candle power 2310, which corre- 
sponds to a specific consumption of 0.295 watt per mean lower 
hemispherical candle power. This curve was the mean of the 
results by three observers, all of whom were in close agreement. 
The sinuosities are probably due to the effect of the globe and the 
reflector and appeared in all the readings. 

Some tests run higher and some lower than the figures here 
given, which may be regarded as a fair average result. The light 
of the quartz-mercury arc is remarkably rich in extreme ultra-violet 


radiations, which are, however, completely cut off by the inclosing 
glass globe, though they make themselves manifest within it by a 
strong smell of ozone. 

There has been some rather unnecessary fear of the quartz arc as 
an illuminant on account of the fact that extreme ultra-violet rays 
are known to react unpleasantly on the skin and particularly on 
the eyes, but the glass globe cuts off these injurious radiations just 
as it does in the case of the magnetite and other powerful electric 
arcs which are also rich in the same radiations, so that, practically, 
the lamp is no more to be feared than any other source of very in- 
tense light and, actually, for a given illumination, delivers as little 
ultra-violet energy as any known illuminant. 

Its high efficiency, great steadiness, and permanency should give 
it a high place among practical illuminants when the quartz tubes 
are more readily obtainable. At the present time they are pro- 
duced only by a few makers, so that the quartz lamp is only 
recently a regular commercial article in this country. 

Finally, one comes to a still different class of gaseous illuminant 
in which the electric discharge takes place in a column of some 
rarefied permanent gas. This type of lamp has often been sug- 
gested but has been produced commercially only in the form of 
the Moore tube, which is in some use and represents an exceedingly 
interesting development of gaseous illuminants. The Moore tube 
is essentially a long Geissler tube fed by an individual high-tension 
transformer coupled directly to the tube so that no high-tension 
wiring is exposed. The tube is generally 1| inches to If inches 
in diameter and in length many feet — up, indeed, to several 
hundred. The tube forms. a closed loop running about the area 
to be illuminated and itself serves practically as the secondary of 
the transformer circuit. 

Fig. 94 shows in diagram the arrangement of the apparatus. 
The transformer is inclosed in a box entered by the terminals of 
the tube. The primary winding is connected to any convenient 
source of alternating-current supply of 60 cycles or so, and in 
series with it is the regulating valve, which is a very essential and 
interesting portion of the apparatus. 

As is well known, the conductivity of a column of gas increases 
rapidly up to a certain point with diminution in pressure. The 
point of maximum efficiency for the Moore tube is about 0.1 
millimeter of mercury, while the maximum conductivity of the gas 



is reached at about 0.08 millimeter pressure. As the tube con- 
tinues in use the pressure in it decreases, and, with no means for 
regulating the vacuum provided, would reach a point which would 
gradually put the tube out of action. To avoid this difficulty a 
branch tube leading to both sides of the lighting tube is turned 
upward at the end and inclosed by a slender conical plug of porous 
carbon, which in the normal action of the tube is just covered by 
mercury. A small solenoid in series with the primary circuit is 

IV — ■ K 

(\ X__^^ \ X MOORE TUB E/ 3 

Fig. 94. 

provided with a core which carries a glass displacing tube, which 
with the normal current in the lighting tube rests in equilibrium 
with the tip of the carbon cone just covered. When the vacuum 
falls below 0.1 millimeter and approaches the critical point of 
conductivity, there is a slight increase of the current through the 
tube, which lifts the displacer, uncovers the tip of the cone, and 
lets gas filter in until the normal vacuum is restored. 

The gas thus fed can be ordinary air when the tube is left open, 
or any convenient gas, a supply of which may be connected with 
the inlet tube of the valve. Ordinary air gives a slightly pinkish 
light; nitrogen, which gives a higher efficiency and is more com- 
monly used, gives a more yellowish tint; and when a nearly white 
light is desired jthe gas employed is CO2. The color given by the 
CQ2 tube in fact is a pretty close approximation to white, con- 
siderably bluer than direct sunlight, but less blue than the light 
of a bright-blue sky. 

The efficiency of the tube as a light producer has been sub- 
ject to considerable study. It is a mistake to suppose that the 
light given by an electrical discharge through gases is necessarily 
efficient; in fact it varies enormously in efficiency according to the 


particular characteristics of the spectrum given by the gas. The 
nitrogen tube, which is that most commonly employed, has been 
found by several experimenters to give a specific consumption of 
approximately 2.4 watts per mean spherical candle power; while 
the CO2 tube as used, especially for color-matching purposes, 
generally in short* lengths, has a specific consumption under these 
conditions of at least 6 or 8 watts per candle. The nitrogen tube 
has, therefore, a slightly greater actual efficiency than the tan- 
talum lamp and materially less than the tungsten lamp. The 
intrinsic brilliancy of the tube, about 0.4 to 0.5 candle power per 
square inch, and the even distribution given by its great extent, 
are practical considerations which tell in its favor. The life of 
the tubes, barring accidents, is very long, running certainly to 
many thousand hours; indeed, it is a little uncertain what except 
accident would limit the life, although in time the interior surface 
of the tube would probably be affected. Very recently tubes filled 
with the rare gas neon have been tried abroad. The color is a 
beautiful orange and the specific consumption is stated to be about 
0.8 watt per candle power, a figure far lower than with any other 
gas yet tried. 

This closes the story of practical illuminants operating by the 
electrical discharge. The list of those unmentioned here would be 
a long and somewhat interesting one, but unprofitable to recite 
from the standpoint of practical illumination. 



As has already been pointed out, the illuminants in common 
use leave much to be desired in the distribution of light, and have, 
for the most part, too great intrinsic brilliancy. The eye may 
suffer from their use, and even if this does not occur the illu- 
mination derived from them is less useful than if the intrinsic 
brilliancy were reduced. 

Hence the frequent use of shades and reflectors in manifold 
forms. Properly speaking, shades are intended to modify the 
light by being placed between it and the eye, while reflectors are 
primarily designed to modify the distribution of the light rather 
than its intensity. Practically, the two classes often merge into 
each other or are combined in various ways. 

Figs. 95 and 96. — Cut-glass Stalactite and Globe. 

There is, besides, a considerable class of shades of alleged deco- 
rative qualities, which neither redistribute the light in any useful 
manner nor shield the eye to any material degree. Most of them 
are hopelessly Philistine, and have no aesthetic relation to any 
known scheme of interior decoration. Figs. 95 and 96, a stalac- 
tite and globe, respectively, of elaborately cut glass, are excellent 
examples of things to be shunned. Cut glass is not at its best 
when viewed by transmitted light, and neither diffuses nor distrib- 
utes the light to any advantage. Such fixtures logically belong 
over an onyx bar inlaid with silver dollars, and to that class 


of decoration in general. Almost equally bad are shades that 
produce a strongly streaked or mottled appearance, like Figs. 97 
and 98. These neither stop the glare from a too intense radiant 
nor render the illumination more practically useful by improving 
its distribution. These shades happen to be for incandescent 
lamps, but they are evil in both principle and application, and 
would be equally bad in connection with any other kind of 

With open gas flames a shade may be of some use as a protection 
from draughts, but generally its purpose is to improve the illumina- 
tion, and if it fails of this it has no excuse for being. For artistic 
reasons it is sometimes even desirable to reduce the illumination 
to a deep mellow glow quit* irrespective of economy, and in such 
case shades may be made ornamental to any degree and of any 

Pigs. 97 and 98. — ShadeB to Avoid. 

density required, or lights may be distributed for purely decorative 
purposes, but gaudy spotted and striped affairs, like those just 
shown, are useless even for these ends. If for decorative purposes 
economy is deliberately set aside, the honest decorator will say 
so frankly. There is no excuse, however, for selling a man shades 
or fixtures certain to double his lighting bill if he tries to get an 
adequate amount of light, while keeping him in ignorance of their 

The' first requirement of a shade is that it shall actually soften 
and diffuse the light it shelters. If it does not do this, no amount 
of ornamentation can make it tolerable from an aesthetic stand- 
point. Almost any kind of ornamentation is permissible that 
does not defeat this well-defined object. Translucent porcelain, 
ground and etched glass, are all available in graceful forms. If 
perfectly plain shades, like Fig. 99, seem too severe, then those 
finely etched in inconspicuous figures, like Figs. 100 and 101, may 


answer the purpose. In such shades the shape is purely a matter 
of taste, subject always to the requirement that the bright source 
must be hidden from all probable points of view. The main 
thing is to conceal the glaring incandescent filament or mantle so 
that it will not show offensively bright spots. Hence the general 
objection to cut glass, which, if used at all, should for the display 
of its intrinsic beauty be so arranged that it can be seen by strong 
reflected light rather than by that which comes from its interior. 
Thin paper and fabrics may be most effectively employed for 
shades and can readily be made to harmonize with any style of 
ornamentation or color scheme that may be in hand. In this 
respect such materials are far preferable to glass or porcelain, 
although more perishable and less convenient for permanent use 
on a large scale. They also entail much loss of light, and are far 
better suited to domestic illumination than to larger installations. 

Figs. 99, 100, and 101. — Shades. 

The real proportion of light cut off by decorative shades has 
not, to the author's knowledge, ever been accurately measured, 
and, indeed, by reason of the immense variety in them, it would 
be almost impossible to average. It is safe to say, however, that 
it is generally over 50 per cent, although the light is so much 
softened that the loss is not seriously felt in reading or in other 
occupations which do not tax the eyes severely. 

With respect to porcelain and glass shades the proportion of 
light absorbed has been measured many times, and on many 
different kinds of shades, so that actual, even if diverse, figures 
are available. The following table gives the general results ob- 
tained by several experimenters on the absorption of various 
kinds of globes, especially with reference to arc lights: 

Clear glass 10 

Alabaster glass 15 

Opaline glass 20-10 

Ground glass 25-30 

Opal glass 25-60 



ilky gloss 30-60 


The great variations to which these absorptions are subject are 
evident enough from these figures. They mean, in the rough, 
that clean clear-glass globes absorb about 10 per cent of the light, 
and that opalescent and other translucent glasses absorb from 15 
to 60 per cent, according to their density. Too much importance 
should not be attached to this large absorption, since it has 
already been shown that in most cases, so far as useful effect 
is concerned, diffusion and the resulting lessening of the intrinsic 
brilliancy are cheaply bought even at the cost of pretty heavy 
loss in total luminous radiation. 

The classes of shades commonly used for incandescent lamps 
and gas lights have been recently investigated with considerable 
care by Mr. W. L. Smith, to whom the author is indebted for 
some very interesting data on this subject. 

The experiments covered more than twenty varieties of shades 
and reflectors, and both the absorption and the redistribution 
of light were investigated. One group of results obtained from 
6-inch spherical globes, intended to diffuse the light somewhat 
generally, was as follows, giving figures comparable with those 
just quoted: 

Per cent. ' 

Ground glass 24.4 

Prismatic glass 20.7 

Opal glass 32.2 

Opaline glass 23.0 

The prismatic globe in question was of clear glass, but with 
prismatic longitudinal grooves, while the opal and opaline globes 
were of medium density only. 

Etched glass, like Figs. 100 and 101, has considerably more 
absorption than any of the above, the etching being optically 
equivalent to coarse and dense grinding. Their diffusion is less 
homogeneous than that given by ordinary grinding, so that they 
may fairly be said to be undesirable where efficiency has to be 
seriously considered. 

A plain, slender canary stalactite behaved like the globes as 
respects distribution, and showed just the same absorption as the 
ground-glass globe, i.e., 24.4 per cent, but permitted an offensively 
brilliant view of the filament within. 

Another group of tests had to do with reflecting shades designed 
to throw light downward, in some cases giving a certain amount 
of transmitted light, in others being really opaque. The char- 



acteristics of some common forms of such shades are plainly shown 
by the curves of light distribution made with the shades in place. 
Pigs. 102 and 103 show two thoroughly typical examples of these 
shades. Fig. 102 is the ordinary enameled tin 8-inch shade, green 

Fig. 102. — Conical Shade. 

Kg. 103. — Fluted Cone. 

on the outside and brilliant white within, a form too often used 
over desks. Fig. 103 is almost as common, being a fluted porcelain 
6-inch shade, used in about the same way as Fig. 102. Figs. 104 
and 105 give the respective vertical distributions produced by 

Fig. 104. — Distribution from 
Fig. 102. 

Fig. 105. — Distribution from 
Fig. 103. 

these two shades, the outer circles showing for reference the, nomi- 
nal 16-ca*dle-power rating. The porcelain not only gives a more 
uniform reflection downwards, but transmits some useful light 
outwards. The case as between it and the tin shade of Figs. 102 
and 104, which gives a strong but narrow cone of light downward, 
may be tabulated as follows: 

8-inch Tin 

Mean spherical candle power 8. 12 

Maximum candle power 29. 49 

Horizontal candle power 0.00 

Absorption, per cent 28. 1 

6-inch Fluted 




The absorption is, of course, based, as elsewhere, on the mean 
spherical candle power. Of these two shades the porcelain one is 
considerably the better for practical purposes. Although it gives a 
somewhat smaller maximum candle power directly below the lamp, 
it gives a much larger well-lighted area, and is for every reason 
to be preferred. A still better form of shade is a plain opal-glass 
cone flashed with green glass on the outer surface. The unaltered 
vertical distribution of an incandescent lamp is given in the curve 
shown in Fig. 56, p. 124, and that curve was from the same lamp 
used in testing these shades. 

It should be noted that the relations of these two forms would 
not be materially altered if they were of appropriate size and were 
applied to Welsbach burners, the distribution of light from which 
bears a rather striking resemblance to that from an incandescent 

Kg. 106.— Shallow Cone. Fig. 107. — McCreary Shade. 

lamp. The tin shade gives too much the effect of a bright spot 
to be really useful for most purposes. If such a concentrated beam 
is desired, it is far better obtained by other and more perfect 

Figs. 106 and 107 show two other forms of reflecting shade in 
somewhat common use, the former designed to give the light a 
general downward direction, the latter to produce a strong and 
uniform downward beam. Fig. 106 is a 6-inch fluted porcelain 
shallow cone) while Fig. 107 is the well-known and excellent 
McCreary shade, 7-inch. They are intended for widely different 
purposes, which come out clearly in the curves of distribution, 
Figs. 108 and 109. 

The flat porcelain cone, Fig. 108, merely gathers a considerable 
amount of light that would ordinarily be thrown upward, and scat- 
ters it outwards and downwards. It has a generally good effect 



in conserving the light, and whether applied to an incandescent 
lamp or a Welsbach deflects downward a good amount of useful 
illumination, but is objectionable in that it does not hide the lamp. 
All the rather flat so-called " distributing " shades should generally 
be shunned for this reason. 

The McCreary shade, on the other hand, is deliberately intended 
to give a rather concentrated beam, softened, however, by the 
ground-glass bottom of the shade. As Fig. 109 shows, it accom- 
plishes this result quite effectively, giving a powerful and uniform 

Vertical on 0° Horizontal 

Figs. 108 and 109. — Curves of Distribution. 

downward beam. The annexed table shows in a striking manner 
the difference in the two cases: 

Flat Porce- 
lain Cone. 

Mean spherical candle power 9.84 

Maximum candle power 15.72 

Horizontal candle power 13.94 

Absorption, per cent 12.8 




The small nominal absorption in the first instance is merely due 
to the fact that the shade is not reached by any considerable por- 
tion of the light, while the large absorption in the later case only 
indicates that nearly the whole body of light is gathered by reflec- 
tion, and sent out through a diffusing screen. 

The porcelain cone is irremediably ugly, but a less offensive shade 
having the same general properties may sometimes be put to a 
useful purpose. The McCreary shade is purely utilitarian, but 


neat, and does its work well in producing a strong, directed illu- 
mination — a bit too concentrated, perhaps, for ordinary desk 
work, for which it should be fitted with a lamp of 8 or 10 candle 
power, but very useful for work requiring unusually bright light. 
Of fancy shades modified in various ways there are a myriad, 
usually less good than the examples here shown. 

In cases where concentration of light downwards along the axis 
of the lamp is . desirable, rather efficient results are attained by 
combining lamp and reflector, that is, by shaping the bulb of the 
lamp itself so that when the part of it nearest the socket is silvered 
on thq outside it shall form an effective reflector of proper shape. 
Obviously when the lamp burns out or grows dim the whole com- 
bination becomes useless, in which respect the device is less 
economical than an ordinary lamp in a carefully designed reflect- 
ing shade like the McCreary. On the other hand, the reflector 
lamps are, on the whole, somewhat more efficient during their 
useful life, and for general purposes of illumination are much less 

In such lamps the bulb, instead of being pear-shaped, is spherical 
or spheroidal, with the upper hemisphere silvered, the silvering 
being protected by a coat of lacquer. The filament usually has 
several convolutions of rather small radius, so as to bring as large 
a proportion of the incandescent filament as possible near to the 
center of the bulb. A filament so disposed throws an unusual pro- 
portion of the light upwards and downwards when the lamp is 
mounted with its axis vertical, but, of course, at the expense of 
the horizontal illumination. 

For various illuminants shades require to be somewhat modified 
in form, and an enormous variety of shades and reflectors are on 
the market, of which those here described may serve merely as 
samples. Shading the radiant, whatever it may be, is a simple 
matter, and so is the use of a pure reflector to direct the light in 
any particular direction. But the commonest fault of powerful 
radiants, as we have already seen, is too great intrinsic brilliancy, 
which calls for diffusion, and good diffusion without great loss of 
light is difficult of attainment, particularly if at the same time 
there is need of redistributing the light so as to strengthen the 
illumination in any particular direction. 

By far the most successful solution of this troublesome problem 
is found in the so-called holophane globes, devised a few years ago 


by MM. Blondel and Psaroudaki, and now in extensive use both 
here and abroad. The general principle employed by these physi- 
cists was to construct a shade of glass so grooved horizontally as to 
form the whole shade of annular prisms. These are not formed as 
in a lighthouse lens, to act entirely by refraction, because in the 
attempt to bend the rays through a large angle by refraction alone 
there is a large loss. 

The prisms of the holophane globe are relieved, as it were, at 
certain points, so that rays which need to be but little deflected 
are merely refracted into the proper direction, while those that 
must be greatly bent to insure the proper direction are affected by 

Fig. 110. — Section of Holophane GIoIjo. 

total reflection. This combination of refracting and reflecting 
prisms in the same structure accomplished the efficient redistribu- 
tion of the light in a very perfect manner. The diffusion remained 
to be effected, and the means adopted was to form the interior 
of the globe into a series of rather fine, deep, rounded, longitudinal 

The total result is a great reduction of the intrinsic brilliancy, 
coupled with almost any sort of distribution required, the total loss 
of light meanwhile being less than in any other known form of 
diffusing shade or reflector. Fig. 1 10 shows in detail, considerably 
magnified, the structure of the holophane prisms and the combina- 
tion of refraction and reflection that is their characteristic feature. 
Here the ray A is merely refracted in the ordinary way, emerging 



with a strong downward deflection from the prism face in the direc- 
tion A 1 . Ray BB l is totally reflected at the face ft 1 , and then 
refracted outwards at 6. C is strongly refracted and emerges from 
the surface c, while DD l is refracted at entrance, totally reflected 
at d 1 , and again refracted at emergence from d. 

The net result is to keep in this particular form of prism surface 
nearly all the rays turned downward below the horizontal. Obvi- 
ously other prismatic forms might be employed, which would give 
a very different final distribution, but the principles involved are 
the same. 

Fig. Ill shows, likewise on a greatly enlarged scale, the interior 
fluting which accomplishes the necessary diffusion of light. The 

Fig. 111. — Diffusing Curves of Holophane. 

ray a is here split up into a reflected component, afterwards re- 
fracted — by e, /, g, and a purely refracted component, 6, c, d. 
The shape of the flutings is such as by this means to secure excel- 
lent diffusion at a very small total loss of light. The inner and 
outer groovings, being at right angles, produce a somewhat tessel- 
lated appearance, but aside from this the surface is quite uniformly 

These holophane globes are made for all kinds of radiants, 
but are most commonly applied to Welsbach gas burners and to 
incandescent electric lights. Evidently the shape of both grooves 
and globe must vary with the purpose for which the shade is 
desired, which results in a very large number of forms, from which 
a selection may be made for almost any variety of illumination. 

It should be noted that these holophane shades both diffuse and 


redistribute the light in a very thorough manner. Speaking gen- 
erally, they are of three distinct classes. The first is laid out 
according to the general principles of Fig. 110, and is intended to 
direct most of the light downwards, serving the same end as a 
reflector, but giving at the same time some needful diffusion with- 
out the use of a ground or frosted globe. The general results are 
strikingly shown in Fig. 112, which gives a graphic idea of what 
such a globe actually docs. 

The second class of globes has for its purpose a fairly uniform 
distribution of the light, mainly below the horizontal, and it is 

Fig. 112. — Holophane, Downward Distribution. 

intended for ordinary indoor lighting, where a particularly strong 
light in any one direction is needless. Its effect is shown in 
Fig. 113. The third general form of holophane globe is designed 
for the especial purpose of throwing a strong light out in a nearly 
horizontal direction, and is shaped so as thus to redistribute the 
light, putting it where it is most useful for such work as street 
lighting, large interiors, and the like. The effect produced is 
admirably shown by Fig. 114. The shapes of globes shown in 
these last three figures are those intended for mantle burners. 

In general, the device enables a good degree of diffusion to be 
secured together with almost any peculiarity of distribution that 


could be wanted, and with a degree of efficiency unexcelled by 
any known system of shades or reflectors, unless it be the Fresnel 
lighthouse lenses. 

Fig. 113. — Holophane, General Distribution. 

One does not generally get such a combination of good qualities 
without certain disadvantages that must be taken in partial com- 
pensation. In the holophane system the weak point is dirt. The 
doubly grooved surface makes an excellent dust catcher, and a 
layer of dust can easily be accumulated quite sufficient to cut down 
the efficiency very seriously. And, moreover, a hasty dab with a 

Fig. 114. — Holophane, Outward Distribution. 

rag does not clean a holophane globe; it must be gone over care- 
fully and thoroughly. When kept clean, the globes actually will 
do just what is claimed for them, and are not at all a merely 
theoretical development excellent only on paper, but they must 
be kept clean, and should not be used where they cannot or will 
not receive proper attention. 


This is probably the chief reason, aside from the extra cost, why 
such globes have not been more extensively used for Street light- 
ing, to which their power of redistributing the light in the most 
useful direction admirably fits them. The results obtained in 
tests of these globes are so striking as to merit examination in 
some detail. 

In spite of the trouble from dust, the holophane globes have 
come into considerable use for street lighting in some European 
citie3, notably Munich, where several thousand have been used on 
Welsbach street lamps for several years past. The net results are 
reported to be exceedingly good, although the amount of labor 
involved must be, from an American standpoint, large. Breakage 
in this case is reported at about 10 per cent per annum. 

If this device could be successfully applied to arc lamps for 
street lighting, a very valuable redistribution of the light might be 
effected, but certain obstacles seem to be interposed on account 
of the shifting of the arc as the carbons are consumed. With a 
focusing form of lamp this trouble would be averted, but such 
lamps have been little used here until the recent advent of the 
flame and luminous arcs which give fumes likely to be trouble- 
some. With inclosed arcs, however, it should be possible to use 
such globes with fair success. 

More recently an interesting modification of the holophane idea 
has been applied to the construction of prismatic reflectors, which 
have come into very large use. The prismatic reflector is essen- 
tially a somewhat bowl-shaped structure of clear pressed glass, 
smooth on the inside, and on the outside formed into a series of 
right-angled prisms, running longitudinally. These prisms act by 
total reflection, returning the light that falls upon them inward 
and downward, so that the reflector acts as though it were pro- 
vided with an exceedingly good reflecting surface. 

Fig. 115, which shows a section of the wall of such a prismatic 
reflector, exhibits the prismatic action. The ray A passes through 
the clear inner surface, strikes one of the outer prismatic surfaces 
at B, is totally reflected to C, and then is totally reflected again, 
and is thus bent back on itself. If the incidence at B were in the 
plane of the paper, the emergent ray would be parallel with the 
entering one. As the rays from the source are generally not in 
what corresponds to this plane, the emergent ray is usually shifted 
downward so that it passes out of the reflector. Obviously, by 



changing the shape of the prismatic reflector this shifting of the 
rays can be controlled so as to modify the distribution. It thus 
is possible to duplicate the types of distribution already shown for 
the holophane globes, while using a reflector open at the mouth 
and thus relieved of any absorption of such light as would natu- 
rally pass freely out of its aperture. 

The advantage gained by this arrangement is the reduction by 
a material amount of the dust difficulty to which reference has 
been made. The interior of the reflector is smooth and does not 
collect dust freely. The dust falling on the exterior surface is 
not in optical contact with the glass, and hence does not inter- 
fere with the total reflections. Such a reflector is not, however, 
completely opaque when viewed from the outside, but diffuses a 






a / 









> > 



Kg. 115. — Showing Principle of Prismatic. 

moderate amount of light which usefully illuminates the surround- 
ing space. This is chiefly due to the fact that the angles of the 
prisms cannot be made absolutely sharp, and consequently rays 
which strike the apex of the prism as D, or the junctions of two 
prisms as F, do not strike at any totally reflecting angle and hence 
pass through as E and G respectively. When more, and more 
uniform, diffusion is wanted than is readily provided in this 
manner, the exterior or interior surfaces can be very finely etched 
or covered with a film of enamel. In such case a large part of 
the light is still totally reflected, but the proportion passing 
through is materially increased, so that the reflector has a soft 
diffusing surface while yet serving as an efficient reflector. This 
diffusion is ordinarily secured by a very delicate acid etching of 
the exterior surface, and many reflectors of this so-called " satin- 
finished" type are in use. They are preferable to the plain re- 


flectors, where the reflector is in full view, since the latter, as 
shown by the diagram, tends to show its diffusion in rather bright 
streaks along the angles of the prisms. 

The best way of examining the performance of these or other 
reflectors is to put a lamp attached to a flexible cord in the 
reflector and have an assistant hold it and awing it slightly from 
side to side while it is under observation from a distance of a few 
yards. It is then easy to see from the appearance of the reflect- 
ing surface whether the light is being widely scattered, moderately 
concentrated, or thrown in a solid beam something after the man- 
ner of a searchlight. It is easy to see in a general way how 
much light is coming through the reflector if of prismatic or opal 
glass and what the distribution of this light is. 

Fig. 117. — Character of Distribution. 

In trying this experiment it will always be found that much 
depends on the exact position of the lamp in the reflector, as can 
readily be told by holding the reflector in one hand and the lamp 
in the other in thus exhibiting it, and then moving the lamp 
axially back and forth. In general, raising the lamp in its re- 
flector tends to concentrate the light, lowering it tends to scatter 
the light. To take advantage of this fact, there are two distinct 
types of shade holders in ordinary commercial use, one of them 
holding the shade high and the other dropping it a little. If a 
particular distribution is desired, it can often be obtained merely 
by the use of one or another of these types. 

The holophane prismatic reflectors are made in three general 
types corresponding to three general types of distribution, — wide, 
medium, and narrow angles respectively. Fig. 116 shows a recent 
type of the first named, intended for tungsten lamps, and Fig. 117 
shows the character of its distribution curve, which tends to spread 



the light rather widely. This so-called " extensive " reflector is 
of service where moderate lighting of a considerable area is under- 
taken with a few lamps. The exact form of the distribution curve 
varies somewhat with the arrangement of the filament and the 
position of the lamp in the reflector, but in a general way the 
maximum illumination is thrown at an angle of nearly 45 degrees 
downward, and the light in this direction is fully double the rated 
horizontal candle power of the lamp. 

Fig. 118 shows the so-called "intensive" or medium-angle pris- 
matic-glass reflector, and Fig. 119 its typical distribution. This 
form is a most generally useful reflector for ordinary cases of 
illumination. It covers, with considerable increase over the rated 

Fig. 118. — Intensive or Medium Fig. 119. — Typical Distribution 

Angle Prismatic Reflector. of Intensive Reflector. 

candle power of the lamp, an angle of from 60 to 90 degrees and 
in a general way gives through this angle 1.25 to 1.5 of the rated 
horizontal candle power. Reflectors having a medium angle of 
distribution should be used in the great majority of practical 
cases where fairly strong lighting is required with average heights 
of ceiling. 

Now and then cases arise in which especially strong local light- 
ing is required, or the lights have to be placed farther than usual, 
from the working plane. In such instances the narrow-angle type' 
of reflector is immensely useful. Fig. 120 shows the so-called 
focusing type of prismatic glass, and Fig. 121 its distribution. 
In this particular reflector the candle power in the axis of the 
beam and its immediate vicinity is about three to four times the 
rated horizontal candle power of the lamp. To meet the most 


extreme conditions, prismatic-glass shades can be obtained which 
will give as high as six or seven times the rated horizontal candle 
power immediately in the axis. These of course are only fit for 
special uses, but sometimes are remarkably convenient. 

With respect to these and all other forms of the holophane 
prismatic glass, it is not safe to assume that the distribution 
curve will rigorously follow the forms here shown except for the 
particular type of shade and the particular lamp with which it 
was tested. In case it is desirable to know the distribution curve 
accurately, it should be ascertained either from the makers or by 
trial for the particular combination of lamp and shade intended to 

Fig. 120.— Focusing T^pe of Pris- Fig. 121. —Character of Distribution 
jntitic Gloss. Focusing Type Reflector. 

be used. There are many varieties of these shades, so that almost 
any required curve can be hit by a little judicious selection. 

Aside from the prismatic glass, there are many good types of 
bowl-shaped reflectors made of opal and similar glasses which 
diffuse a moderate amount of light and reflect the rest at fair 
efficiency from the interior surface. The best of these give a per- 
formance quite similar to that shown in Fig. 119, and they do good 
work in cases where medium-angle distribution is required. They 
do not, however, so effectively give either a wide-angle distribution 
or a very narrow-angle one. For obtaining a concentrated beam, 
some mirror reflectors and reflectors with a polished interior sur- 
face give results quite similar to those in Fig. 121. If the surface 
of these is made slightly matt and carefully shaped, a fairly con- 
centrated beam can be obtained without the scattering reflections 



from the filament which usually appear in a highly polished sur- 
face. Fig. 122 shows such a curve derived from an approximately 
parabolic steel reflector finished with aluminum on the inside and 
intended for use with a 25-watt tungsten lamp. It will be ob- 
served that the candle power in the axis rises to nearly five times 
the rating of the lamp. Reflectors of this description are of great 
service in workshop lighting, where lamps must often be suspended 


150° 1B0° ISO 

Fig. 122. 

on drop cords and where the shades occasionally may get hard 

One cannot leave this subject of reflectors without mentioning 
an interesting and occasionally useful type in which, while the 
lamp is axially situated in the shade, the distribution of the light 
is unsymmetrical, so that if the shade be used on a bracket the 
light is thrown out from the wall instead of against it. This result 
is obtained in prismatic-glass reflectors by an ingenious combina- 
tion of the original holophane structure with the totally reflecting 


prisma already mentioned. Fig. 123 shows an asymmetric shade 
of this type, and Fig. 124 its very curious distribution of light. 
In Fig. 123 the left-hand half of the shade is of ordinary holophane 
construction, while the right-hand half is composed of totally 
reflecting prisms, with the result that more than two-thirds of 
the light is thrown to one side of the shade, as shown in Fig. 124. 
At 45 to 50 degrees below the horizontal on that side the avail- 
able candle power rises to about one and a half times the rating 
of the lamp. Such asymmetric shades are manufactured in a 
number of forms, and it is unnecessary to state that to secure 
their proper operation great care must be taken to see that the 
portion containing the reflecting prisms is turned away from the 
direction which it is desired to illuminate. Somewhat similar 
effects can be obtained with opal glass or with metallic reflectors 

Fig. 123. Fig. 124. 

by cutting away a portion of one side of the shade and forming 
it accordingly, but these are open to the objection of exposing 
the lamp, which is generally undesirable. 

The addition of a diffusing coating to a shade of any description 
somewhat tends to round the distribution curve, so that this treat- 
ment affects unfavorably very wide or very narrow angle reflectors, 
but is nevertheless occasionally desirable. 

Finally, we must pass to a group of shades and reflectors of a 
highly specialized character, used for lighting walls and ceilings 
while concealing the source of light wholly or sometimes partially 
from direct view. These are employed either for lighting special 
things like pictures or bookshelves or for indirect general illumina- 
tion wherein none or very little of the light which reaches the eye 
is derived from the radiant but chiefly from light diffused by walls 
or ceiling. Reflectors of the first class are practically troughs of 
section specialized for the work in hand. They are commonly 


made of metal, usually employ lamps with their axes parallel to 
the length of the trough and well hidden by it, and may be lined 
for reflecting purposes either with mirror strips or with a brilliant 
interior coating of some description. 

A very good example of such a device is shown in Fig. 125. 
This was designed for lighting bookcases in a library, is placed 
above and a little in front of the surface 
to be illuminated, and contains, prefer- 
ably, tubular lamps in the position shown. 
The whole interior has a polished reflect- 
ing surface, and the direct rays of the 
lamp are cut off by the reflecting half- 
cylinder, in the axis of which it lies. The 
curve is specially designed to give uniform 
illumination over a considerable surface, 
and the group of tangents to the various 
points of the surface suggests at once the 
obvious way of laying out a reflector for 
such service. Indeed, the practical way 
of designing such reflectors is to start with 
the desired sheaf of rays necessary to illu- 
minate the surface required and trace 
these back to an assumed position of the 
source, passing the reflector curve in the 
simplest possible way through the loci 
defined by the tangents derived from g ' 

tracing back the rays. By following out this scheme, beautifully 
uniform illumination can be secured, particularly if the surface 
is slightly matt to avoid strong direct reflections of the fila- 
ment. These reflecting troughs take a multitude of forms ac- 
cording to requirements, and are of considerable use in practical 

To a different category belong the devices intended deliberately 
for the indirect illumination of rooms. They belong in general to 
three types. The earliest of these is the reflecting cove, which is 
a curved cornice made structurally a part of the finish, and con- 
taining a recess in which can be placed lamps backed by a suitable 
reflector. The light from these lamps illuminates the curved 
plaster or painted surface of the cove above and contiguous por- 
tions of the ceiling into which the eove fades away. There is 



therefore produced a brilliantly illuminated cornice which serves 
as the secondary source of radiation for the illumination of the 




Fig. 126. 

room. Such a cove has a cross section similar to that shown in 
Fig. 126. Its kinship to the curved reflector of Fig. 125 is obvious, 
but the object in the case of the cove is not to provide uniform 


illumination over a wide area, but somewhat brilliant illumination 
over a comparatively restricted area, so that the curves are slightly 
different. The efficiency of this and other schemes for indirect 
lighting will be taken up in its appropriate place. 

A second group of indirect-lighting fixtures are practically in- 
verted reflectors throwing their light wholly or chiefly upon the 
ceiling, which then serves as a secondary source of light. Obvi- 
ously, prismatic glass or metallic reflectors such as are used in the 
ordinary way can be readily applied to such indirect illumination, 

Kg. 127. 

the former providing more or less diffuse light, the latter concealing 
the source entirely. 

The inverted arcs with metallic reflectors have been freely used 
for some years past in this way, but it is only with the advent of 
the metallic filament lamps that attention has been drawn to 
indirect illumination of this class with incandescent sources. One 
of the typical forms of indirect-lighting fixture for such use is 
shown in Fig. 127. This is a so-called " X-ray " reflector of corru- 
gated glass silvered on the surface and then protected from tar- 
nishing by a coat of elastic enamel. The reflector proper is carried 


in a metallic casing with a suitable fixture provided for holding the 
lamp in the axis of the reflector, and the whole may be mounted 
on a fixture or suspended from chains so that the light which would 
otherwise fall below the horizontal is thrown entirely toward the 
ceiling. Of course it is necessary that the reflecting surface should 
be kept reasonably free from dust, in this as in all cases of reflectors 
for indirect lighting. The device, however, has come into consider- 
able use as a convenient way of securing indirect illumination with- 
out special structural provisions in the building. 

Finally, a group of fixtures should be mentioned which are es- 
pecially designed to accomplish the same result that could be reached 
by using a prismatic glass reflector inverted; that is, they are so 
arranged as to throw a cpnsiderable part of the light upon the 
ceiling for indirect lighting and at the same time to diffuse a soft 
illumination through the space below. These direct-indirect or 
semi-indirect fixtures, as they are sometimes called, have great 
artistic possibilities and can be made to give beautiful illumina- 
tion, but they have not yet come into large use, although the 
scheme is an old one. 

All these methods of indirect and semidirect lighting are rela- 
tively inefficient, and not time enough has yet elapsed since the 
general introduction of high-efficiency incandescent lamps and 
mantles to develop the auxiliary appliances to their full measure 
of usefulness. 



The lighting of houses is a most interesting and generally neg- 
lected branch of illumination. Artificial light has been distinctly 
a luxury until within comparatively recent times, and in domestic 
lighting there has not been the same pressure of commercial neces- 
sity which has resulted in the general efforts to illuminate other 
buildings. Indeed, until within half a century there was very little 
effort at really good illumination in the home, everyone depending 
on portable lights, which could be brought directly to bear upon the 
work in hand; gas, which provides fixed radiant points, being con- 
fined to large cities, and in these to houses of the better class. Even 
at the present time very little pains is taken to arrange the lighting 
in a systematic and efficient manner. 

The comparatively small areas to be lighted in dwellings, the 
small need for extremely intense light, and the very discontinuous 
character of the need for any light at all, render domestic lighting 
rather a problem by itself. Of ordinary illuminants all may be 
freely used for such work, save arc lamps and very powerful 
gas lamps, such as the large regenerative burners and the most 
powerful incandescent mantles. 

Arcs are of very unnecessary power, hence most uneconomical, 
and are often so unsteady as to be most trying to the eyes. 
In the home, as a general thing, one does not keep the eyes 
fixed in any definite direction, as one would if working steadily 
by artificial light, so that far more than usual care must be 
taken to avoid intense and glaring lights. Therefore, arcs are 
highly objectionable, and the gas lights of high candle power 
equally so, particularly as the latter throw out a prodigious 
amount of heat and burn out the oxygen of the air rather 

As to other illuminants, the main point is to choose those of 
low intrinsic brilliancy, or to keep down the intrinsic brilliancy 
by adroit and thorough shading. Anything over two or three 
candle power per square inch it is well to avoid as needlessly 



trying to the eyes without any compensating advantage save 
economy, which can better be secured in other ways. 

Aside from the physiological side of the matter, very bright 
lights seldom give good artistic results or show an interior at 
anything like its true value. Of the common illuminants, gas 
and incandescent lamps are those generally most useful, while 
petroleum lamps and candles are even now auxiliaries by no means 
to be despised. Professor Elihu Thomson once very shrewdly 
remarked to the writer that if electric lights had been in use for 
centuries and the candle had been just invented, it would be hailed 
as one of the greatest blessings of the century, on the ground 
that it is absolutely self-contained, always ready for use, and 
perfectly mobile. 

Now, gas and incandescents, while possessing many virtues, lack 
that of mobility. They are practically fixed where the builder or 
contractor found it most convenient to install them, for while 
tubes or wires can be led from the fixtures to any points desired, 
these straggling adjuncts are sometimes out of order, often in 
the way, and always unsightly. Besides, the outlets are often for 
structural reasons in inconvenient locations, and their positions 
need to be chosen very carefully if artistic effects are at all to be 
considered; so that while these lights are the ordinary basis of 
illumination wherever they are available, lamps and candles, which 
can be put where they are wanted and not necessarily where some 
irresponsible workman chose to locate them, are often most useful 
additions to our resources. 

In domestic, as in other varieties of interior illumination, two 
courses are open to the designer. In the first place, he can plan to 
have the whole space to be lighted brought uniformly, or with 
some approximation to uniformity, above a certain brilliancy, 
more or less approximating the effect of a room receiving daylight 
through its windows. Or, throwing aside any purpose to simulate 
daylight in intensity or distribution, he can put artificial light 
simply where it is needed, merely furnishing such a ground- 
work of general illumination as will serve the ends of art and 

While the first method is for purely utilitarian purposes often 
necessary, it is frequently uneconomical and inartistic in its 
results. Its sin against economy is in furnishing a great deal 
of light which is not really needed, while in so doing it usually 


sends light in directions where it deadens shadows, blurs contrasts, 
and illuminates objects on all sides but the right one. The second 
method is the one uniformly to be chosen for domestic lighting, 
from every point of view. 

In electric lighting the most strenuous efforts are constantly 
being made to improve the efficiency of the incandescent lamps 
by a few per cent, and an assured gain of even 10 per cent would 
be heralded by such a fanfare of advertising as has not been heard 
since the early days of the art. Yet in lighting generally, and 
domestic lighting in particular, a little skill and tact in using the 
lights we now have can effect an economy far greater than all the 
material improvements of the last twenty years. The fundamental 
rule of putting light where it is most useful, and concentrating it 
only where it is needed, is one too often forgotten or unknown. 
If borne in mind it not only reduces the cost of illumination, but 
improves its effect. 

In applying this rule in practice, one of the first things which 
forces itself upon the attention is the fact that the conditions can 
seldom be met by the consistent use of lights of one uniform 
intensity, or one uniform characteristic as regards the distribution 
of the light around the radiant. Even one kind of illuminant 
is sometimes an embarrassing condition. Both the kind and 
quantity of the illumination must be adjusted to the actual 
requirements, if real efficiency is to be secured. 

As has already been shown, the effective illumination depends 
upon two factors, — the actual power of the radiant in candles or 
other units, and the nature of the surroundings, which determine 
the character and amount of the diffuse reflection which reenforces 
the direct light. If the radiant in a closed space furnishes a certain 
quantity of light, L, then the strength of the illumination pro- 
duced at any point within the space will depend, if the walls are 
nonreflecting, simply on the amount of light received from the 
radiant, in accordance with the law of inverse squares. If the walls 
reflect, then the total illumination at any point will be that received 
directly, L, and in addition a certain amount kL (where k is the 
coefficient of reflection), once reflected, a further amount k 2 L twice 
reflected, and so forth. The total illuminative effect will then be: 

L(l+A; + fc 2 + fc 8 + . . . k n ). 

As k is obviously always less than unity, this series is convergent 


upon the limiting value L( _ , J, which expresses the relative 

effect of the walls in reenforcing the light directly received from 
the radiant. 

It is clear from the values of k already given for various sur- 
faces that such assistance may be of very great practical import- 
ance. A simple experiment showing the value of the light diffusely 
reflected is to read at some little distance from the radiant in a 
room having light walls, and then to cut off the direct rays by a 
screen close to the radiant and just large enough to shade the book. 
If the conditions are favorable, the amount of diffused illumination 
will be somewhat startling. A repetition of the experiment in a 
room with dark walls will exhibit the reverse condition in a most 
striking manner. 

A good idea of the practical amount of help received from dif- 
fusion may be gained by computing the effect for various values 
of k. The following table shows the results for values of k between 
0.05 and 0.95: 

k L 


0.95 20.00 

.90 10.00 

.85 6.66 

.80 5.00 

.75 4.00 

.70 3.33 

.65 • 2.85 

.60 2.50 

.55 2.22 

.50 2.00 

.45 1.81 

.40 1.66 

.35 1.53 

.30 1.42 

.25 1.33 

.20 1.25 

.15 1.17 

.10 1.11 

.05 1.05 

In practice the interior finish of dwelling houses is highly hetero- 
geneous, the walls being tinted and broken with doors and hang- 
ings, the ceiling being often of another color, and the floors covered 
with colored rugs or carpets, and generally provided with furniture 
at least as dark as the walls. The floor is in point of fact the least 
important surface from the standpoint of illumination, for it not 


only carries the furniture, but from its position cannot diffuse light 
directly in any useful direction. So far as it is concerned, only 
the small terms in k % and higher powers enter the general equation, 
since the illumination diffused from below is not of much account. 

These values show the great difference between good and poor 
diffusing surfaces in their practical effect. Reference to the table 
already given shows that ordinary wall surfaces give values of k 
ranging from about 0.60 down to 0.10 or less. These are likely to 
be reduced by the gradual absorption of dust at the surface, but 
it is quite within bounds to say that the effective illumination in 
a room may be nearly or quite doubled by the light diffused from 
the walls. If an average value of fc is computed on the basis of 
the respective areas and values of k for the several surfaces of the 
room, the above table gives in practice a pretty accurate idea of the 
reenforcement of the direct illumination. 

The ceiling is a very important consideration, for the light 
diffused downward is highly valuable. Vaulted ceilings are notori- 
ous in their bad effect upon the illumination. If used at all, they 
should be employed with full knowledge of the fact that they quite 
effectively nullify all attempts at brilliant general illumination, and 
when considerations of harmony permit, ceilings ought to be very 
lightly tinted. 

As to the walls themselves, wainscoting and dark soft-finished 
papers absorb light very strongly, and render lighting difficult, 
while the white-painted wood and light papers freely used in 
Colonial houses produce exactly the reverse effect. The character 
of interior finish, being determined by the contemporaneous fashion, 
can of course seldom be really subordinated to the matter of illu- 
mination, which affects only personal comfort; but in planning a 
scheme of decoration it is necessary to bear in mind that the darker 
the general effect the more light should be provided 

The outlets for Jas and electricity provided for and quite ade- 
quate to light a brightly finished house, will prove entirely insuffi- 
cient if a scheme of decoration in dark colors be afterward carried 
out, so that it is the part of wisdom to arrange the original outlets 
to meet the worst probable conditions for lighting. This will gener- 
ally mean arranging for about double the minimum amount of 
illumination necessary on the hypothesis of strong diffusion from 
the walls. 

If conditions demand or fashion dictates any attempt at very 



bright illumination, a sort of simulated daylight, all matters relat- 
ing to diffusion are of very serious import. Fortunately, such is 
not the usual case. Where the main purpose is that already strongly 
urged, — of merely furnishing such illumination as is necessary for 
practical or artistic purposes, — there need be no effort at uniform 
intensity of light or at making dark corners brilliant; and, while 
the aid of favorable diffusion is still important in reducing the total 
amount of artificial light furnished, it no longer so completely con- 
trols the situation. 

With the data now at hand we can form a fairly definite idea 
of the quantity of light which must generally be provided. One 
can get at the approximate facts by considering the amount of 

Kg. 128. — Vertical Section of Room. 

light that must be furnished in a room of given size to bring the 
general illumination up to a certain value. The particular value 
assumed must depend upon the purpose for which the room is to 
be lighted. For instance, since 1 foot-candle is an amount which 
enables one to read fairly well, let us assume that we are to fur- 
nish, in a room 20 feet square and, say, 10 feel high, a minimum 
of 1 foot-candle. 

To start with, we must make some assumption as to the amount 
gained by diffusion from ceiling and walls. For this, in a con- 
crete case, we can make an educated guess from the data already 
given. In general, Wybauw found that in moderate-sized rooms 
the diffusion increased the effective value of the radiant 50 per 
cent, which, as it agrees pretty closely with our own values, taking 
into account a light ceiling, we will use for the present purpose. 



Let the assumed radiant be at r, Fig. 128, and at a height of 6 
feet 6 inches above the floor. Now draw an imaginary plane ab 
at a height of 2 feet 6 inches above the floor, and take this as the 
surface to be illuminated. If r is in the center of the room, the 
g reate st distance from r to a corner of the plane ab will be 
V216 feet = 14.7 feet. Each candle power at r must be reduced 
proportionately, so that 1 candle at r would give *H foot-candle 
at the point in question. According to our hypothesis, diffusion 
aids by 50 per cent, so that instead of requiring 216 candle power 
to give 1 foot-candle in the remotest corner, the real amount 
would be 144 candle power, which would be handily furnished by 
a cluster of nine 16-c.p. incandescent lamps or their equivalent. 
The result would be a room quite brilliantly lighted, for, except 






< io< > 









\ A 


Hg. 129. —Floor Plan. 





N V 


Fig. 130. — Floor Plan. 

very near the walls, the illumination would be much in excess of 1 
foot-candle, rising to 4 or 5 foot-candles upon the plane of lighting 
under and near the lights. 

Such an arrangement of the lights is, however, uneconomical 
in the extreme, since the distant corners are illuminated at a very 
great disadvantage. Fig. 129 shows the advantage gained by a 
rearrangement. Here the room is divided by imaginary lines into 
four 10-foot squares, and in the center of each of these is a light 
6 feet 6 inches above the floor, as before. Now, if a corner of the 
plane of lighting, as E, receives 1 foot-candle, the requirements 
are fulfilled. But E is distant from D just about 8 feet, from 
C and B almost exactly 16 feet, and from A less than 22 feet. 
It, therefore, receives, neglecting A, for each candle power at 
D fa foot-candle, and for each at C and B a total of T i* foot- 


candle, or, allowing for diffusion, J$ and ^V respectively (nearly), 
so that it at once becomes evident that four 32-c.p. lamps are more 
than sufficient to do the work. 

Taking A into account, four 25-c.p. lamps would almost suffice, 
but obviously the maximum illumination is perceptibly lowered. 
It would be a maximum at the center, and for 32-c.p. lamps 
would there amount to 2 foot-candles. A still further subdivision 
would lead to still better distribution from the point of view of 
economy, and, indeed, something can still be gained by a further 
redistribution of the light; for, with lights arranged as in Fig. 130, 
at the center and on the circle inscribed in the room in question, 
five 20-c.p. lamps would very closely fulfill the conditions, reducing 
the total amount of light required to meet the assumed condition 
from 144 to 100 candle power in all. 

Obviously, with a fixed minimum illumination and no other re- 
quirement, the conditions of economy will be most closely met by 
a nearly uniform distribution of the minimum intensity required. 
There is, however, a limit to practical subdivision in limited areas, 
such as rooms. In the case of large buildings, as we shall pres- 
ently see, one can easily figure out the illumination on the basis 
just taken, but in domestic lighting we have to deal with a very 
limited number of radiants, at least in considering gas and elec- 

By far the best results are attained by providing a very mod- 
erate general illumination and then superposing upon it strong 
local illumination for special purposes. For example, in most 
rooms better practical results than those of Fig. 130 would be 
reached by following the same arrangement, but using four 
16-c.p. or even four 8-c.p. lamps and one 32-c.p. lamp, the latter 
being placed near the point where the strongest illumination is 
required. The result would be to give the extreme corners all 
the light they really need, and to provide plenty of light where 
it is of most practical value. In ordinary domestic lighting the 
four smaller lights would often be put on brackets and the large 
one installed in a table lamp. 

The same rules apply to the use of gas or other illuminants, 
always bearing in mind that the total amount of light required 
is strongly affected by the hue of the walls, and that the principal 
radiant should be placed where it will do the most good. Illu- 
mination thus regulated is both safer physiologically and far more 


efficient in use of the material than any attempt at uniform distri- 
bution over the entire area. 

One's choice of illuminants must obviously be governed by the 
question of availability. Incandescent electric lamps easily hold 
the first place when economy is not the first consideration, by 
reason of their being quite steady, giving out little heat, and in 
no way vitiating the atmosphere. They should always, however, 
be furnished with ground bulbs, or, better, so shaded as greatly 
to reduce their otherwise very high intrinsic brilliancy. 

Next in order of desirability unquestionably comes gas. Used 
with the incandescent mantle burner, it is the cheapest known 
illuminant for domestic purposes unless electricity can be ob- 
tained at exceptionally low rates. Mantle burners should always 
be shaded, both to reduce the intrinsic brilliancy and to modify 
the hue of the light, unless some of the recent mantles giving an 
amber tone to the light are available. Ordinary gas jets, in case 
of need, give a good but expensive subordinate illumination. 

Lamps and candles have strong merits for particular purposes, 
but are inferior for general work. The former are often used 
with good effect to furnish the principal radiant, which may be', 
reenforced by small gas lights. Candles, on the other hand, are 
extremely useful for partial and subsidiary illumination, since they 
are the only available source of small intensity unless one goes 
to considerable trouble in wiring for tiny electric bulbs, which are 
better adapted to purely decorative purposes than to the regular 
work of illumination. 

From this general basis of facts we can now take up the prac- 
tical and concrete side of domestic lighting. 

As to the distribution of the lights required for interior illu- 
mination, one must be guided by the intensity which is necessary. 
The examples already given show the general character of the 
problem. The laws upon which the solution depend may be 
formulated as follows: If we write L for the required or existing 
intensity of illumination in foot-candles at any point, C for the 
candle power of the radiant, and d for the distance in feet from 
that radiant, then: 

L-% C-L* 


If the point in question receives light from more than one radiant, 


the illuminative effects must be summed, and, if the radiants are 


L is of course in foot-candles and C in candle power. In 

these expressions no account is taken of the varying angles of 

incidence of the light received from the several radiants. In 

C cos i 
principle, L = — -= — , where i is the angle of incidence; in other 

words, the illumination decreases as it becomes oblique. 

In certain cases account must be taken of this fact, but since, 
as a rule, objects to be lighted are oblique to the plane of illumi- 
nation, and cos i is small only in case of rather distant lights, of 
which the entire effect is small, and since the diffused light cannot 
be reckoned with, having no determinate direction, the question 
of obliquity, particularly when the radiants are numerous and well 
distributed, has seldom to be dealt with. It is rendered the more 
uncertain by the notorious inequality of the distribution of the 
light from ordinary illuminants, and it must be remembered that 
the whole aspect of the matter is changed by the use of reflectors. 
It is better to take the obliquity factor by general average in 
assuming the illumination required. 

In ordinary interior illumination one constantly meets limita- 
tions imposed by structural or artistic considerations. For example, 
we have already seen that the arrangement shown in Fig. 130 was 
highly desirable for economic reasons. The five lamps dangling by 
cords or rods, from the ceiling of a room 20 feet square, might be 
tolerated in an office, but would be quite inadmissible in a drawing 
room. For domestic lighting one is more likely to use chandeliers, 
side brackets, and ceiling lights. The last-named have been con- 
siderably used of late, sometimes with beautiful effects, sometimes 

To examine the effect of ceiling lights on the situation, refer 
to Fig. 131, which shows the same room as Fig. 128. Assuming 
the same general conditions, let us find the illumination at a point 
p in the plane of illumination when given by a light r in the old 



position, and a ceiling light r', 6 inches below the ceiling. The light 

being assumed as of 16 candle power, the light at p is L = jt = 0.39 

1 fi 
foot-candle, when the lamp is at r, or L = =j = 0.21 when the 

lamp is at r', close to the ceiling, neglecting diffused light. 

In a room very bright with white paint or paper, having, for 

example, k = 0.60 and f 1 _ , J = 2.50, the total illumination would 

be 0.39 + 0.97 = 1.36, and since the diffusion does not materially 
change with the position of the light, the illumination in the second 
case would be, roughly, 0.21 + 0.97 = 1.18; in other words, the 

Fig. 131. — Location of Ceiling Lights. 

change in position of the light would make but a small change in 
the intensity of the illumination. 

There is evidently some error made in assuming that diffusion 
increases the illumination by a certain ratio, and Wybauw's hy- 
pothesis of replacing the diffused light by an imaginary radiant 
directly above the real radiant involves the same error. It is prob- 
ably nearer the truth to assume, in case of an apartment having 
several radiants, that the total illumination at any point is that 
due to the lights severally, plus a uniform illumination, due to 
diffusion and proportional to k and C. 

The practical upshot of the matter, however one may theorize 
on the rather hazy data, is that shifting the lights in a room from 
their usual height to the ceiling does not affect the illumination 
seriously if the walls and ceiling diffuse strongly, while if they are 
dark the change is decidedly unfavorable. This does not, however, 


imply that ceiling lights should not be used in dark-finished rooms, 
although it is very plain that if they are so used the lamps should 
be provided with reflectors, or themselves form reflectors, as in 
some lamps recently introduced. 

If the walls have a very low coefficient of diffusion it is obvious 
that all light falling upon them is nearly wasted, at least from the 
standpoint of illumination, and therefore the economic procedure 
is to deflect this light so that instead of falling upon the walls it 
shall be directed upon the plane of illumination, which is chosen 
to represent the average height from the floor at which are the 
things to be illuminated. If reflectors or their equivalents are 
skillfully applied, the radiants, for the purpose in hand, are nearly 
or quite doubled in intensity, so that there is a good opportunity 
for efficient lighting. But these reflecting media must be used 
with caution to avoid the appearance of beams giving definite 
bright areas, and by far the best results may be obtained by using 
diffusing shades in every such case. So far as economy of light 
is concerned, reflectors can be advantageously used wherever the 
effective reflection exceeds the total diffusion coefficient of the walls. 
For example, with a hemispherical reflector having a coefficient of 
reflection of 0.70, the hemispherical intensity of the radiant is 
1.70 C, assuming a spherical distribution of the light. This value 
corresponds, so far as the plane of illumination is concerned, with 
a diffusion of k = 0.40, which signifies that, except in very light 
finished rooms, the radiant is used more efficiently by employing a 
reflector than by trusting to the really very serviceable diffusion 

from the walls. 

The use of side lights close to the 
wall, or on short brackets, is preferable 
to lighting from the ceiling in certain 
cases, as when strong local illumination 
is needed. Reflector lamps may here 
again be used with very great effect if 
the walls are at all dark in tone. Fig. 
132 gives in diagram the simplest ar- 
rangement of such lamps. We may 

t* ,oo Q .j t • u* assume their height as a trifle less than 

Fig. 132. — Side Lights. . *T 

in the case of the suspended lights, 
say, 3 feet above the plane of illumination, and that they are 
equipped with reflectors giving a hemispherical distribution of 


light. In Fig. 132 the positions of the lamps are indicated by 
black dots, as before. It is evident that the corners will be the 
points of minimum illumination, and that in the central part 
of the room the lighting will be rather weak, although, on the 
whole, the distribution of light will be good. With help from 
diffusion to the extent assumed in the last example, four 20-c.p. 
lamps would do the work, while with dark walls the case would 
call for at least four 32-c.p. lamps. In fact with dark walls 
lighting from brackets becomes extremely inefficient. 

Now, summarizing our tentative arrangements of light, it appears 
that to illuminate a room 20 feet square and 10 feet high on the 
basis of an approximate minimum of 1 foot-candle will require from 
80 to 144 effective candle power, according to the arrangement of 
the lights, if the finish is light, and half as much again, at least, if 
the finish is dark. The floor space being 400 square feet, it appears 
that the illumination is on the basis of about 3 to 5 square feet per 
effective candle power. The former figure will give good illumina- 
tion under all ordinary conditions; the latter demands a combi- 
nation of light finish and very skillfully arranged lights. 

For very brilliant effects, no more than 2 square feet per candle 
should be allowed, while if economy is an object, 1 candle power 
to 4 square feet will furnish a very good groundwork of illumina- 
tion, to be strengthened locally by a drop-light or reading lamp. 
The intensity thus deduced we may compare to advantage with 
the results obtained by various investigators, reducing them all to 
such terms as will apply to the assumed room which we have had 
under discussion. 

Just deduced 1 c.p. per 3 sq. ft. 

Uppenborn 1 c.p. per 3.6 sq. ft. 

Piazzoli 1 c.p. per 3 . 5 sq. ft. 

Fontaine 1 c.p. per 7.0 sq. ft. (approximation). 

In very high rooms the illumination just indicated must be 
somewhat increased, owing to the usual necessity for placing the 
lamps rather higher than in the case just given, and on account 
of the lessened aid received from diffuse reflection. The amount 
of this increase is rather uncertain, but in very high rooms it would 
be wise to allow certainly 1 candle power for every 2 square feet, 
and sometimes, as in ballrooms and other special cases requiring 
the most brilliant lighting, as much as 1 candle power per square 


On the other hand, in most domestic lighting, the amount of 
lighting needed may be reduced by a little tact. Ordinary living 
rooms, such as parlors, libraries, and the like, do not require to 
be uniformly and brightly lighted in most cases. It is quite 
sufficient if there is ample light throughout the main portion of 
the room. 

A groundwork illumination of 0.5 foot-candle over the whole 
room, plus a working illumination of 1.5 to 2 foot-candles in addi- 
tion over a part of the room, gives an excellent result. This is 
something the result that would be reached in Fig. 130 by using 
a 32-c.p. central lamp and four 10-c.p. lamps for the rest of the 
room. Dining rooms need ample light upon the table, but do 
not in the least require illumination of equal power in the remote 
corners. Sleeping and dressing rooms do not require strong light 
so much as well-placed light. A bedroom of the dimensions we 
have been discussing could be very effectively lighted with three 
or four 16-c.p. lamps, provided they were placed where they would 
do the most good. 

To go into detail a little, perhaps the most important rule for 
domestic lighting is never to use, indoors, an incandescent or other 
brilliant light, unshaded. Ground or frosted bulbs are of much ser- 
vice when incandescents are used, and opal shades, or holophane 
globes, which still better reduce the intrinsic brilliancy, are availa- 
ble with almost any kind of radiant. Ornamental shades of tinted 
glass or of fabrics are exceedingly useful now and then, when ar- 
ranged to harmonize with their surroundings. 

In incandescent lighting the lamps may be placed in any posi- 
tion. With gas or other flame radiants ceiling lights are not 
practicable, although inverted Welsbachs may be raised fairly near 
the ceiling. As to the intensity of the individual radiants, con- 
siderable latitude may be given. In many instances, incandescents 
or gas or other lights of as low as 8 to 10 candle power are con- 
venient, while for stronger illumination radiants of 15 to 20 candle 
power reduce the cost of installation, and for special purposes lights 
of 30 to 50 candle power, incandescents or incandescent gas lamps, 
are most useful. 

At the present time the general introduction of the tungsten 
lamp and of the small mantle burner has made the task of efficient 
domestic lighting very much easier than ever before. At the price 
customarily charged for current in house lighting, using carbon 


filament lamps is simply a waste of money without any conceiv- 
able benefit by way of excuse. 

The only difficulty with the tungsten lamp is its fragility, which 
is still a serious matter in spite of reputed and real improvements. 
The danger of accidental breakage from blows or jarring is the 
one source of annoyance in using these lamps in house lighting. 
Barring accident, their life is sufficiently long to make it well 
worth while to use them even at the present high price, but since 
in house lighting there is very seldom any necessity for using the 
larger sizes of tungsten lamps, the fragility of the slender filaments 
found in the smaller sizes must be taken into serious consideration. 
It is very rarely that one wishes to use a lamp larger than 40 
watts in domestic lighting. Most of the work falls to the 25-watt 
size, and a very considerable proportion of lamps for house lighting 
will be of still smaller size, — 15 watts or thereabouts, when these 
lamps are available, as they even now are at reduced voltage. 

These small tungsten lamps should be generally either operated 
by wall switches or provided with pendent switches or pull switches. 
Key sockets should be employed with caution for the tungsten 
lamps of 25 watts and below, not so much on account of the 
danger from the snap of the switch as for the risk of fumbling 
around the socket and accidentally hitting the fixture hard enough 
to break the lamp. The writer, however, uses a good many 
15-watt 55-volt tungsten lamps in his own house op. key sockets 
with very little trouble from breakage, but these low-voltage lamps 
are somewhat sturdier than usual, and it is questionable whether 
the same immunity from trouble would have been experienced in 
using 110-volt lamps of this small size. There is no reason why 
a large proportion of lamps in a house should not be upon 
switches. It will here be well to take up systematically the lighting 
of a house, considering carefully both the amount of light required 
and the way in which it is advisable to apply it. 

The lighting of a house has a very intimate relation to the 
decorative effect sought. Indeed, all the difficulties of the situa- 
tion are due to the concessions which have been made to the 
decorative situation. Unfortunately there are no fixed canons of 
taste regarding interior decoration. It is, on the contrary, almost 
purely a matter of fashion, without the slightest philosophic basis; 
consequently, interiors which would have been considered charm- 
ing fifty years ago may be decried as execrable at the present 


time, and again lauded as the ideal of decorative art fifty years 
hence. At least such is the history of the subject in the past, 
which there is every reason to expect will be repeated. In plan- 
ning the illumination of a house, therefore, one must be prepared 
to meet occasionally conditions that render effective lighting well- 
nigh impossible, as well as others in which it becomes extremely 

The first task of the illuminating engineer is to provide a suffi- 
cient number and capacity of outlets to give good illumination in 
spite of all the subsequent efforts of the decorator. These outlets 
need not be used to their full capacity, but they should be avail- 
able in case of necessity. The four usual methods of lighting a 
room are as follows: ceiling lights, pendent lights, either in chan- 
deliers or on flexible cords, brackets, and table lamps. The choice 
between one of these devices and another is determined by the 
character of the room to be illuminated. Just at the present 
moment decorators, who are an imitative folk, are booming the 
bracket. Now, the bracket in small or exceptionally narrow rooms 
of light finish can give excellent results in illumination, but if the 
wall finish is dark the effectiveness of the lamp placed near it 
suffers greatly, and it is not an easy matter to throw light out into 
the room from a bracket without the use of shades or reflectors 
of very special design. Moreover, there is a constant temptation, 
from a slavish following of old precedents, to make a bracket 
simulate a sconce, or at least so to design it as to require its equip- 
ment with candle lamps. From the standpoint of the illuminating 
engineer the candle lamp is something to be avoided, not only 
on account of the obvious objection that it is at best a shabby 
counterfeit of a candle, but from the fact that candle lamps are 
highly special as an article of manufacture, inefficient, and difficult 
to shade adequately without cutting off an objectionably large 
proportion of the light. 

Some of the most unsuccessful and inartistic lighting which the 
writer has ever had the misfortune to see has been by the mis- 
application of brackets as the only source of light to rooms with 
extremely dark wall finish. The use of brackets in this way is 
historically inharmonious, since the scheme belongs to a period 
of civilization in which either very little light was required, — far 
less than can suit even the most modest modern requirements, — 
or when the brackets were used as auxiliaries to chandeliers, which 


furnished the major part of the lighting. This combination of 
chandelier with a multitude of glittering candles, and candles used 
as side lights to reenf orce it, is capable of giving altogether charm- 
ing results; but the side brackets alone, fitted with imitation 
candles, are both ineffective and out of keeping with the situation 
in which they are generally found. If one wants sconces or the 
like as bric-a-brac, he should at least have the courage of his 
convictions and fit them with genuine candles. 

Chandeliers can be made both beautiful in themselves and 
effective, but they belong properly in large and stately rooms, 
high enough to give them place without breaking up the contin- 
uity of the room, and big enough to allow a chandelier on a suf- 
ficient scale to give it its historic decorative character. Like the 
bracket, it belongs to special rather than to general conditions. 
Lamps placed close to the ceiling either singly or in clusters form 
essentially a modern type of lighting fixtures, since they could not 
have been used successfully prior to the introduction of electric 
lights. They are, therefore, too frequently held in horror by the 
decorator, although in the hands of some of the skillful fixture 
designers at the present time very graceful and harmonious exam- 
ples have been turned out. For domestic use they are extremely 
useful in small- and moderate-sized rooms, and there is assuredly 
no logical reason why a fixture, in itself well designed to meet 
the conditions imposed by modern illuminants, should not be 
artistically every bit as good as any possible adaptation of such 
illuminants to fixtures which belong characteristically to a differ- 
ent period and to methods of illumination now obsolete. A fixture 
is not necessarily good merely because it is a somewhat slavish 
imitation of a seventeenth- or eighteenth-century model. 

Beading lamps of one kind or another have a history as ancient 
as could be desired. They are, moreover, available in many 
forms, suitable to almost every possible requirement, and should 
occupy in modern domestic lighting a considerably more promi- 
nent place than is generally accorded to them. Their installation 
requires a liberal supply of wall or baseboard plugs and floor 
receptacles if they are to be successfully adapted to electric light- 
ing. There is no reason why similar taps should not be applied 
to gas lighting, although the usual form of the gas reading lamp 
is dependent for its supply on piping from a gas chandelier. 

A combination of table lamps with brackets or any other com- 


mon means of lighting can be made effective from every point of 
view. Sometimes, though rarely, the lighting can be trusted to 
lamps alone. The writer, for example, found it desirable to illu- 
minate his own library in this way, the room being long and 
low, so that its continuity would have been hopelessly broken by 
ceiling lights or chandeliers, and lined with bookcases to an extent 
that precluded the effective use of brackets. The use of lamps on 
this scale is not particularly economical, owing to the character of 
the shades which generally have to be used, but otherwise it leaves 
little to be desired. 

Finally, one may mention various plans for indirect lighting, 
such as have already been described. The design of artistic fix- 
tures for this class of work has not progressed far enough to 
warrant one in being enthusiastic about it, and the concealed 
cornice lighting, while it may be made very effective, is objec- 
tionable in the house on account of extreme difficulty in keeping 
the reflecting surfaces clean. It therefore is a form of illumina- 
tion which should be used with caution. 

A word here concerning the matter of fixtures, whatever type 
of illumination be attempted. The fact must be recognized that 
fixtures may have a high decorative value if properly designed. 
With this phase of the matter the illuminating engineer is not 
directly concerned, except in so far as bad design may interfere 
with the suitability of the fixtures for the purpose of their use, 
which is to supply light. The best fixture designers fully recog- 
nize this, and can, if given a free hand, produce fixtures which are 
excellent from every point of view. From the standpoint of illu- 
mination, the chief thing to be borne in mind is that the fixture 
must not be so located or so formed as seriously to interfere with 
its illuminating function. The writer has in mind one case in 
which some very elaborate wrought-iron chandeliers were pro- 
vided with deep bell-shaped black iron receptacles within which 
the incandescent lamps were located and almost wholly concealed. 
After divers ineffective efforts to secure a perceptible amount of 
light from these, the unfortunate owner resorted to "gas arcs" of 
the most glaring and undecorative description. 

The most effective way of protecting a client against bad fix- 
tures is to persuade him to cut down his fixture appropriation 
to a point that will enforce simplicity and unobtrusiveness. 
The worst fixtures, as a rule, are the somewhat showy ones of 


medium price in which an attempt has been made to obtain a 
so-called decorative effect without the skill in design, and finish in 
execution, really necessary for good results. 

If one wishes to make fixtures a distinct decorative feature in 
an interior, he must be prepared to pay for the privilege, and so 
far as illumination is concerned its employment for a purely dec- 
orative object is quite legitimate. When so used the man who 
pays the bills should clearly understand that he is using light as a 
decorative element, quite irrespective of its primary use for illu- 
mination, and must also understand that he is decorating rather 
than illuminating his room by such use of lights. If he is willing 
to pay for it on this basis, that is his business, and no concern of 
the illuminating engineer. 

The writer recalls a case in his own practice where a private 
library was subjected to an electric light bill about ten dollars a 
month in excess of what should have been required under ordinary 
illuminating practice, merely on account of the nature of the in- 
terior finish and the wholesale use of brackets with candle lamps. 
The owner grumbled at his electric light bills, although he prob- 
ably received florists' bills of much larger amount, also incurred for 
decorative purposes, with entire equanimity. 

Some attention has here been paid to this phase of the matter 
because the illuminating engineer and decorator are popularly 
supposed to be at swords 1 points. Such is not at all the case in 
reality, since the illuminating engineer is not concerned with the 
taste of his client in the selection of fixtures or bric-a-brac. He 
should, however, make it plain that if he has to work under cer- 
tain fixed limitations regarding decorative effect, he cannot be 
expected to give good illumination for the purpose of seeing with- 
out correspondingly large cost to the owner. And, on the other 
hand, there should be no concealment of the fact that the decorator 
may be advising the use of finish and fixtures that will double or 
even quadruple the cost of lighting a room for the ordinary pur- 
pose of its use. 

Taking up now the lighting of a house in detail, one may 
summarize the situation about as follows: 

A. Halls. — The illumination of these depends entirely on the 
way in which they are to be used. Ordinary entrance halls can 
generally be sufficiently lighted on the basis of 4 or 5 square feet 
per rated candle power. Elaborate halls, which are likely to be 


used considerably in entertaining, as part of the working space, 
should have a larger allowance of light, say 1 candle power to every 
2 or 3 square feet. The basis of reckoning is here nearly inde- 
pendent of the height of the ceiling for the reason that dwelling 
houses do not commonly present any extremes of height requiring 
special consideration. Back halls and other subsidiary halls require 
less light, perhaps 1 candle power to 6 or 7 square feet, since they 
are used only as passageways. Lanterns and side brackets are 
the usual means selected, and the location should be such as to 
thoroughly light the stairway. With exceptionally low ceilings, 
lanterns are out of scale unless small and located close to the 
ceiling, and brackets will more generally answer the purpose. 
Occasionally one sees very beautiful staircase lighting from a newel 
post, but the scheme is one that had better be avoided except in 
work on a large scale and when the designer can be given carte 
blanche for the necessary fixture. 

B. Reception Rooms. — These are usually of somewhat formal 
character, to which the fixtures must correspond. The amount 
of light required at times is considerable and should be provided 
on the basis of not more than 3 square feet per candle power, 
preferably 2, although the subdivision of units should be such 
as to allow half or one-third of the light only to be used under 
ordinary conditions. One generally finds it advisable in reception 
rooms to use brackets or chandeliers, or frequently a combination 
of the two, the chandelier being really suitable only in rather large 
and high rooms, and better replaced for low ceilings by a ceiling 
fixture. Sometimes side brackets and table lamps make a suitable 
combination, the former for general illumination, the latter for 
added brilliancy when needed. Here, as everywhere in a house, 
it should be the invariable rule of the illuminating engineer never 
to allow a bare incandescent lamp to be visible from any point 
where one is likely to be placed. If bare lamps are to be used at 
all, they should be behind diffusing shades of one sort or another, 
otherwise the lamp should be completely frosted, and even the use 
of frosted lamps should be shunned on account of their generally 
too high surface brilliancy. 

C. Music Rooms. — Music rooms are employed, or are sup- 
posed to be employed, for a definite function. They do not ordi- 
narily require high illumination save near the instrument, which is 
best cared for by a suitable lamp so as to throw the light upon 


it and not into the faces of the audience. Above all things, the 
lighting should be restful in effect and the lights extremely well 
shielded. Unobtrusive chandeliers or ceiling clusters with lights 
well screened are the best means of meeting this requirement. 

Brackets are particularly objectionable in such rooms, inasmuch 
as they are constantly in the field of view and it is rather difficult 
to screen the lights adequately. The amount of light provided 
should be on the basis of 3 to 4 square feet per candle power, the 
latter figure being quite all that is necessary, unless the room is 
to be at times used for general purposes requiring a little more 
illumination. These figures presume a wall finish not excessively 
dark, and any room wainscoted in dark wood, or finished in paper 
or paint in dull reds, greens, browns, and blues, would require not 
less than 50 per cent more 'light than here specified. Now and 
then some form of concealed lighting is effectively used in a music 
room. The main point, however, is that in any room where an 
audience is to be assembled the light must be kept out of their 
eyes, since they are not at liberty to change their positions and 
escape it. 

D. Libraries. — - By tradition libraries are usually given a dark 
finish, and if actually used as libraries the walls are lined with 
bookcases to an extent which implies powerful absorption of light. 
Further, they are supposed to be used for reading, which requires 
fairly good illumination, so that the, amount of light supplied 
must of necessity be considerable. The best way of furnishing 
it is to apply a groundwork illumination, either from fixtures 
near the ceiling or from brackets, in either case thoroughly 
shaded, and then to strengthen it locally by well-shaded reading 

The total amount of light required will be commonly 1 candle 
power for 3 or 4 square feet, about half the total amount being 
put in the general illumination and half in the reading lamps. In 
case of exceptionally dark finish a total of 1 candle power to as 
low as 2 square feet may be desirable. Lamps near the ceiling are 
much to be preferred to brackets for general illumination in a 
library, since with these latter it is almost impossible adequately 
to light the bookcases, which must lie nearly in the same plane 
with the brackets. Further, if a library is really used, it is 
usually very difficult to find any space for wall brackets without 
interfering with the bookcase space. The library should be wired 


for an ample supply of current, floor plugs and baseboard plugs 
being especially useful. 

As a rule fixtures, here as elsewhere, should have the lamps 
pointing either directly up or directly down, preferably the latter. 
Fixtures having a lamp at an angle are not in the least necessary 
for artistic results, and are extremely objectionable from the fact 
that lamps so placed are exceptionally difficult to screen properly. 
One is practically obliged in using such fixtures to employ globes 
completely inclosing them, which waste an unnecessary amount 
of light without gaining important advantages. In ceiling lights 
either inclosing globes or reflecting shades can be employed, which 
can be made adequate to cover the lamp, and can be given 
directive action sufficient to supply good illumination beneath. 

E. Living Rooms. — A living room; used as such, is best lighted 
by a combination of table lamps and general illumination from 
lights placed at or very near the ceiling. The amount required 
is a trifle less than in the ordinary library on account of the usually 
lighter finish. One candle power to 4 square feet, at least half of 
it being in table lamps placed where they will do the most good, 
ought to be adequate. Few living rooms are large enough or 
formal enough in character to make the use of chandeliers desirable. 
Nothing is more garish and generally ineffective than the ordinary 
three- or four-armed chandelier with the lights at an angle and 
placed at a height where they are likely to be hit by unguarded 
motions or the passage of someone more than usually tall. More- 
over, lights in such a place cannot be properly screened except 
by closed globes, and even these are thrust into the face of any- 
body entering the room. If the lights are placed high, at or near 
the ceiling, they are out of the ordinary field of view and can be 
made to put illumination where it will be useful. 

F. Dining Rooms. — The dining room, more perhaps than any 
other room in the ordinary house, is the prey of unthinking and 
irresponsible fashion. Table lamps, except when small and for 
purely decorative functions, are bad, since they obstruct the view 
across the table. Now and then in dining rooms of the formal 
and stately type chandeliers may be used with very beautiful effect, 
but they belong distinctively to very high rooms giving ample 
space for decorative effects, carried well above the immediate view 
of the persons in the room. 

Side brackets, unless for decorative use only and with the lights 


exceedingly well screened, are highly objectionable here, since they 
must be shining directly into someone's face and cannot be escaped. 
With a room too low and too formal to permit a proper chandelier, 
ceiling clusters answer admirably, and from the standpoint of com- 
fort a good deal is to be said for the domes, which at times have 
been popular and may be highly decorative, although at present 
anathema from the ephemeral standpoint of fashion. 

There is much to be. said, too, for concealed and " semi-direct " 
lighting in rooms of this sort, where it is highly undesirable to 
have a glare of light in the faces of those sitting at table. The 
amount of light required is fortunately not great. One candle 
power to 4 or 5 square feet is ample if applied with any intelligence. 

In passing, the writer may say that altogether the most beautiful 
effect he has ever seen in any private dining room was produced 
by the use of an exquisite eighteenth-century chandelier in crystal, 
reenforced by candelabra, all carrying real candles, without any 
aid from modern illuminants. 

G. Kitchen and Pantries. — The service portion of a house in 
general requires about the same illumination found elsewhere, 
with this exception, that in some of the rooms already mentioned 
provision must be made for exceptionally bright lighting on par- 
ticular occasions. The service portion requires merely good work- 
ing illumination at all times. By far the best way of securing this 
is by lights placed practically at the ceiling with suitable prismatic 
glass or similar reflectors. In rooms of ordinary height this is the 
position of the greatest advantage. Lights in a kitchen may be 
located at one or more points as the arrangement of the working 
space requires. 

The total amount of illumination required is not great. In the 
kitchen itself an average of 1 candle power to 5 or 6 square feet is 
ample. Pantries require a little more, being generally active work- 
ing spaces. One light, which sometimes has to be on a bracket, 
although a properly located ceiling light frequently answers, should 
be arranged to give good light in the interior of the ice chest. All 
lights in the service portion should be on wall switches or pull 
sockets, if tungsten lamps are to be used, as is commonly desir- 
able. In lighting the service portion of a house with gas, the same 
provisions for amount of light and general location hold, and it 
pays to use automatic gas lighters, which pay for themselves many 
times over in decreased breakage of mantles. The same is true 


of the use of gas lights through all other portions of the house, 
the rules for amount of light and most advantageous location hold- 
ing rigorously true, irrespective of the particular illuminant used. 

The chief trouble with gas lighting is the supposed necessity of 
getting the lights far enough down to enable them to be lighted 
with a match and the consequent wholesale use of inartistic and 
inconvenient fixtures of the chandelier type. In these days of 
inverted mantle burners and thoroughly- worked-out systems of 
lighting them, there is no difficulty in getting adequate and con- 
venient illumination from gas. The combination gas-and-electric 
fixtures are generally abominations from the artistic standpoint, 
quite unnecessary, and not to be recommended for any purpose. . 

H. Bedrooms. — Bedrooms generally suffer rather from badly 
placed lighting than from inadequate amount. The actual amount 
required is not large, say 1 candle power for every 5 square feet 
in rooms of ordinary finish. These are the only rooms in the ordi- 
nary house in which brackets are positively advantageous, although 
here they generally could be reenforced to advantage, at least in 
electrically lighted houses, by a small ceiling light for purposes of 
general illumination. 

Bureaus and dressing tables can be better lighted from swinging- 
arm brackets than by any other device yet tried, since these 
brackets can be moved to exactly the position where they will be 
of the most service. The one place where a light should never be 
placed is above and a little in front of the bureau or dressing table. 
This is the position often picked out by the thoughtless and igno- 
rant, who act as if the mirror and the top of the bureau were the 
things to be lighted, instead of the person standing or sitting in 
front of them. 

A swinging bracket, placed not too high and carrying a lamp 
of moderate candle power well shaded, is the form of local lighting 
best suited for such use. An additional light placed over or beside 
the bed and easily reached from it is a most useful addition to the 
bedroom lighting equipment. At least one light should be switched 
from the door, and the others may either be switched at the lamps 
or from a point near at hand. Three lighting units are com- 
monly required for the ordinary bed room, more if the room is 
exceptionally large. These units should preferably all be small, 
unless a special lamp be desired for reading purposes. The same 
general conditions hold for lighting with gas. Sometimes a night 


lamp is a desirable addition; with electric lighting, a 2-c.p. lamp is 
excellent for this purpose; with gas lighting, one small jet shielded 
against the danger of being blown out by draughts. 

/. Billiard Rooms. — Where these are found in private houses 
they are generally small, containing only one table. This requires 
very strong lighting, say two lamps with downward, somewhat 
concentrating reflectors, each lamp from 30 to 50 candle power 
and equally spaced over the table area. For electric lighting two 
40-watt or 60-watt tungstens are generally sufficient. With gas 
lighting two small inverted Welsbachs give similar candle power. 
These lights should be placed rather high, so as to be as far as 
possible out of the way of the cue in mass£ shots. There is a good 
deal to be said for the use of indirect lighting in billiard rooms for 
the avoidance of shadows, which, while mostly suppressed by the 
dark hue of the cloth, are sometimes embarrassing. In case this 
method is tried, two 100-watt tungstens or two large Welsbachs, 
with suitable reflectors to throw the light on the white ceiling, are 
none too great an allowance of light for the work. When the 
lighting is direct, care should be exercised in picking out the reflec- 
tors so as to avoid streaks of light from direct reflections of the 
filament. A matt interior surface with ordinary reflectors, or satin- 
finished prismatic glass, would generally meet this requirement. 

J. Basements. — Booms in the basement of a house commonly 
require only a very moderate degree of lighting, varying from 1 
candle power to 4 square feet in the portions most used to half 
this amount elsewhere. The location of the lights is generally 
definitely fixed by the shape of the subdivisions in the basement 
or cellar. Lights generally should be placed as near the ceiling 
as possible. 

K. Bathrooms. — Very little need be said for these. One small 
light is usually sufficient and is best placed on a swinging bracket. 
The bathrooms are usually of such size that the smallest ordinary 
unit gives ample illumination, especially since the finish as a rule is 
very light. 

L. Closets. — Most closets unless extremely well lighted from 
the room require one small light, generally at the ceiling. The 
same is true of storerooms and attic space. It is sometimes good 
policy to place these lights, and particularly those in closets, on 
automatic door switches, so that they will not be left burning 
through carelessness. If only used in one or two places it is com- 


monly unnecessary, but where a considerable number of closets are 
thus fitted it is wise to use automatic switches. 

In general, domestic lighting is peculiar, in that a very large 
number of lamps relatively to the average or maximum load is 
installed. The ordinary dwelling house of ten rooms or above will 
require from forty lights upwards, and the number of outlets, in- 
cluding base plugs, is likely to be half as great again unless rigorous 
economy in the original installation is necessary. The economy 
of domestic lighting depends on the convenient arrangement of 
the lights in such wise that only those necessary at any time shall 
be in use; hence it saves money to install ample and convenient 
switching, or, in case of gas lighting, automatic lighting apparatus. 

Electric lights in the halls, both front and back, and in the cellar, 
should be put on three-way switches, so that they can be lighted 
and extinguished from more than one place. Clusters and chan- 
deliers when used may advantageously be wired in two or more 
circuits, so that except when full illumination is needed a small 
amount of light can be used. In an ordinary electrically lighted 
house at least nine-tenths of the illuminating work can best be done 
by 25-watt and even 15-watt tungstens, very few larger lamps or 
special lamps being ordinarily required. It is advisable, too, in 
providing the wiring, to have several taps of ample capacity, in 
addition to the lighting connections, provided for small heating 
devices, fan motors, and vacuum-cleaner connections. Occasion- 
ally someone objects to the relatively white light of tungsten lamps 
in domestic use, but this can so easily be toned down by the use 
of tinted shades or faint tinting of the lamps themselves by dipping, 
that it constitutes no serious objection; and the efficiency, even 
after the color is modified, lies always with the metallic filament 



Before passing to the interesting problem of lighting large in- 
teriors, it may be well to consider the group of transitional cases 
represented by the rooms of an ordinary office building, and par- 
ticularly the smaller ones. Such buildings are variously arranged, 
presenting rooms of divers sizes, used for offices or small counting 
rooms, and occasionally for the display of goods. As a rule they 
are of moderate size, ranging from 200 to 1000 square feet, and 
commonly run from 11 to 13 feet in height of ceiling. The ordinary 
finish is on the whole thoroughly light, although the wood-work 
may be dark, and in many instances the natural light is rather 
poor, so that artificial lighting has to be resorted to for a con- 
siderable part of the day. The outlets in office buildings are 
generally badly planned, the only redeeming feature being ceiling 
outlets, useful for general illumination. Brackets, when found, are 
generally fixed with the lights pointing downward at an angle of 
45 degrees, by all means the worst possible position, and are chiefly 
useful as attachment places for portable lamps. Suitable base- 
board outlets for adjustable plugs are too often infrequent and 
inconveniently placed. 

Now, in all such rooms there are two radically distinct modes of 
procedure. One can either provide a moderate general illumination 
and reenforce it by portable lamps placed upon desks or tables, or 
one can provide a general working illumination alJ over the room. 
With the scanty outlets commonly provided the former course is 
the easier, and there is no objection to following it, except in rooms 
of the larger class. In these one often sees a veritable network of 
lamp cords, very unsightly, and very much in the way. In addi- 
tion to this disadvantage, it is almost impossible to provide enough 
lights for a considerable group of tables and desks without seriously 
inconveniencing some of the workers through the glare of their 
neighbors' lights. It is sometimes, therefore, highly desirable to 

resort to general illumination, which can be carried out successfully 



and economically if the outlets and desks or other working spaces 
can be conveniently arranged. 

The main difficulty in general illumination is not physical but 
psychological. Those who have habitually worked with badly 
placed and ill-shaded individual lights have found by experience 
that under these conditions a great deal of light is necessary in order 
to enable them to see. The eye is working under very disadvan- 
tageous circumstances, and is dazzled by the glare, which has to be 
reenforced, so to speak, in order to leave enough residual light by 
which to see. Now this experience becomes almost an obsession, 
so that the worker forms a settled opinion that he or she can get 
adequate light for seeing only by using a large lamp equipped with 
a powerful reflector thrust immediately down over the work, and 
draws the further conclusion that any lamp which is not powerful 
and not close to the work cannot give adequate illumination. 
Hence, when an attempt is made to change from individual to 
general lighting, there is almost always a violent complaint that 
the lighting is insufficient, although in point of fact it may be more 
than adequate for the work and very much easier on the eyes than 
the illumination that it has replaced. After some experience in 
working with general illumination the eye gets accustomed to its 
conditions and works with much less effort even with materially 
lower illumination. 

Nearly all classes of clerical and office work can be performed 
easily under an illumination of 3 to 4 foot-candles. In rooms 
where bookkeeping is carried on to a considerable extent the 
larger figure mentioned is about right, while the more ordinary 
office occupations can get along admirably with 3 foot-candles or 
even a little less. The exception is made in favor of bookkeeping 
because men at work on ledgers and filing slips usually write with 
fine pens for economy of space and with ordinary office ink of a 
somewhat bluish cast as it flows from the pen. The combination 
is a bad one from the standpoint of lighting. In cases where 
general illumination is not attempted and individual lamps are 
used in addition to a general groundwork lighting, the latter may 
conveniently be between 1 and 2 foot-candles, with desk lamps 
sufficient to bring the lighting on the work to a point not exceed- 
ing 4 or 5 foot-candles. Running above this is totally unnecessary 
and is apt to be trying for the eyes. Whichever method is adopted, 
all the lights, both those furnishing general illumination and those 


lighting the work for individuals, should be thoroughly screened 
so that they cannot shine into anybody's eyes. 

Lighting really large interiors differs in several important respects 
from ordinary practice as applied to rooms of the medium sizes 
just considered. In the first place, the aid received from diffusion 
from the walls is much less than in the case of smaller rooms, 
as has already been indicated. The experiments of Fontaine 
indicate that within moderate limits the light required is deter- 
mined by the volume of the space to be illuminated, rather than 
by the floor space. 

Since, however, the only physical effect of the increased height 
is to increase the mean distances of the diffusing surfaces and 
especially the ceiling from the radiants, the change could, in point 
of fact, alter ohly that part of the total illumination due to diffused 
light, provided that with increased height of ceiling the radiants 
are not themselves raised. 

In large and high rooms there is a strong tendency to increase 
the height of the radiants above the plane of illumination, especially 
in case of using chandeliers, and this is the most important factor 
in the rule aforesaid. Obviously, in increasing the distance of the 
radiants one decreases the direct illumination approximately in 
the ratio of the inverse squares of the distances, and does not 
materially improve the diffusion. 

Therefore the illumination falls off seriously. In a large and 
high hall, lights arranged in the ceiling or as a frieze, while often 
giving admirable effects, are quite uneconomical, and should be 
used, if at all, with a full appreciation of this fact. 

In large buildings, too, the quantity of light required is subject 
to enormous variation, according to the purposes to which the 
building is devoted, and whether the whole interior must for artistic 
reasons be illuminated. In a ballroom an effect of great brilliancy 
is generally aimed at, while a room of equal size used as a factory 
needs strong illumination only where it will facilitate the work. 

Again, in very large rooms the power of the individual radiants 
can advantageously be increased, and some sources of light in- 
admissible in domestic lighting, such as arc lamps; or to be used 
only with caution, like powerful mantle gas burners may be 
used very freely. 

But in large buildings, as elsewhere, the fundamental purpose 
of the lighting is to produce a certain intensity at the plane of 



illumination, which in such work should be assumed about 3 feet 
above the floor. The absolute illumination required may vary 
greatly, over a range, in fact, as great as from half a foot-candle 
to 3 foot-candles or more; but the lighting may properly be calcu- 
lated from an assumed value, just as in the cases already discussed. 
For purposes of discussion, we may first consider a hall 100 feet 
long by 30 feet high by 50 feet wide. The plane of illumination 
will then have an area of 5000 square feet, and the total volume 
is 150,000 cubic feet. And for simplicity we will assume 1 foot- 
candle as the minimum intensity to be permitted in any part of 
the space. Fig. 133 shows the plan of this assumed space. We 
will first take up the case of suspended radiants, which is the most 
usual method of treating such a problem. 

Fig. 133. — Plan of Hall. 

Obviously, in a room of the shape given a single radiant is out 
of the question, on the ground of economy, since in meeting the 
requirement of a given minimum of illumination the most eco- 
nomical arrangement is that which exceeds this minimum at the 
fewest points possible. Two radiants give a possible solution, and 
are worth a trial. Clearly, they must be located on the major axis 
of the room AB; but since a corner, as E, is the most unfavorable 
spot to light, the radiants must be placed well toward the ends of 
the room. We will assume their height as 15 feet above the floor, 
and 12 feet above the plane of illumination. 

Now the best place for the given radiant a is easily determined: 
it is such that, calling the projections of the points E and C upon 
the plane of illumination E l and C l , aC 1 = aE l y/2> approximately. 
To fulfill this condition Aa = Bb = 15' very nearly, and the two 


radiants are at once located. In this case cP = 994, and since 
C = Ld 2 , C should be practically 1000 candle power. Allowing 

(nh) = 

1.5, each of the radiants should be of about 666 candle 

power, a requirement which could be practically met by a nominal 
2000-candle-power open arc, if its glare were not so forbidding. 

Using incandescents, 42 of 16 candle power would be required 
in each group, which should be increased to about 60 if ground 
bulbs in a chandelier were to be used, since lamps so mounted 
interfere with each other's effectiveness to a certain extent. Re- 
ducing these figures to square feet per candle power, it appears 
that the assumed conditions are satisfied by allowing as a maximum 
about 3.75 square feet per candle power, or with allowance for 
properly softening the light, 2.6 square feet per candle power. 

Lighting such a space from two points only is usually by no 
means the best way, and a much better effect would be secured 
by using six radiants. The same reasoning which led us to place 
a and b near the ends of the major axis of the room indicates a 
similar shifting in the case of six lights. From symmetry, two 
should be on the minor axis DOC, and as regards the projections 
of C and on the plane of illumination, the best position for a 
radiant, located in the same horizontal plane as before, is at a', 
about 6 feet from C, with V at a corresponding point on the other 
side of 0. Now for the lateral pairs of lights. One of them may 
be approximately located with reference to E l , and the projection 
of the middle point of the line to a', much as a' itself was located. 
This leads to a position c', 41 feet from a' and 9 feet from the 
wall. Forming now the equation 

C = —r^ V d * = 306 ' dl% * 1906 ' 

n (iP + *°-) 

and the sum of the other terms is little greater than the term in 
rfi 2 . Simplifying thus, the candle power of each radiant comes 
out very nearly 235, without allowance for diffusion on the one 
hand or for ground bulbs and incidental losses on the other. 

It therefore appears that the conditions call for 15 16-c.p. 
lamps in each of the six groups, a total of 90 as against 120 in 
the previous arrangement. The total rated candle power is 
then 1440, or 1 candle power for every 3.5 square feet. Six 250- 


watt tungstens with suitable reflectors would very likely prove 

It is interesting to check this computation, based entirely on an 
assumed minimum illumination of 1 foot-candle, with the result of 
experiment. For large rooms, ranging from about 1000 to 5000 
square feet in area, Uppenborn's careful investigations show that 
for good illumination 3 to 3.5 square feet per candle power is the 
amount required in practice. In most cases these large spaces are 
finished in light color, so that in spite of the high ceilings they 
are scarcely more difficult to light than ordinary dwellings. The 
absolute brilliancy required is determined by the purpose of the 
illumination, and the proper arrangement of the lights depends 
largely on architectural considerations. Oftentimes frieze and 
ceiling lights are used in halls, and their application to the case 
in hand is worth considering. 

If arranged as a frieze, the lamps might be equally spaced 
around the walls, at about 5 feet below the ceiling, bringing them 
22 feet above the plane of illumination. For simplicity we will 
assume the use of 90 16-c.p. lamps, with reflectors or their equiv- 
alent. Each gives approximately 27 candle power in its hemi- 
sphere of illumination. These lamps would be spaced a little 
more than 3 feet apart, giving 30 on each side of the hall and 15 
on each end. Now, taking for examination the corner E 1 , which 
is as unfavorable a locality as any, and roughly running up the 
illumination at this point, it falls a little short of 1 foot-candle, 
but a diffusion factor of 1.25 would carry it just about to the 
required amount. With lightly ground bulbs, which are far prefer- 
able to the clear ones in such a case, an increase to 36 lamps on 
each side and 18 on each end would be desirable, and 40 and 20 
on sides and ends respectively would do still better. With the 
tungsten lamps now generally used the tendency is toward the use 
of larger units. 

Lighting from the ceiling would lead to a slightly worse result. 
Lights so arranged, however, can give a very valuable groundwork 
of illumination when reenforced by lights more favorably placed. 
They have the advantage of being unobtrusive and of producing 
a generally brilliant effect, but give, if used to the exclusion of 
everything else, an illumination painfully lacking in chiaro-oscuro, 
a difficulty which is keenly felt in some forms of indirect light- 
ing; and light directed almost entirely downwards is, moreover, 


somewhat trying, suggesting a stage scene in the absence o\ foot- 

As has been already explained, the illumination at any par- 
ticular point should have a predominant direction, else the effect 
on the eyes is apt to be annoying. A room lighted by brilliantly 
phosphorescent wall paper, for example, would produce a most 
disagreeable effect unless the luminosity were confined to one side, 
or, in general, to limited portions of wall. 

Something of the same objection appertains to ceiling or frieze 
lighting when pushed to an extreme. In the room under dis- 
cussion, the best general effect would probably be produced by 
combining pendent or bracketed lights with about an equal 
amount of illumination from frieze or ceiling lights. 

Having thus obtained an outline of the lighting of a simple large 
area, we may, before passing to some of the special cases of large 
interiors, profitably take up one which practically is of great im- 
portance — the lighting of extensive and comparatively low rooms. 
This is one of the most frequent tasks which the illuminating 
engineer has to encounter. It is found in stores, in many fac- 
tories and machine shops, in rooms employed for clerical work, 
and to a certain extent in ordinary offices. A typical case may 
be found in the lighting of a room, say 30 by 60 feet and no more 
than 10 to 12 feet in the story. 

Here, then, is a room of some 1800 feet of floor space and of a 
height not much over one-third its width. Windows will usually 
be found only on one side of such a room, sometimes only on one 
end, so that ample provision has to be made for artificial light- 
ing. The ceiling is generally light and the walls as generally rel- 
atively dark. The most typical case of this kind is where the 
space is used for clerical or general office purposes, for which 
strong and even illumination is necessary. The illumination 
required for such a room will be usually from 2.5 to 4.5 foot- 
candles on the working plane, the former figure for general office 
purposes, the latter for clerical work of a more trying kind. 

Almost the first question which arises is the method by which 
the illumination can properly be calculated in a room of this 
kind. Obviously a considerable number of light sources will be 
used. To what extent do they all share in furnishing the illu- 
mination at a given point on the working plane? Some of the 
lights will be shining almost directly down upon the work, others 


at a high angle of obliquity and therefore furnishing, according to 
the cosine law, only a small fraction of illumination. Theoreti- 
cally, one is at liberty to integrate the light received at a given 
point on the working plane from all directions, but practically a 
large part of the oblique light is either cut off or ineffective. On 
the other hand, it does contribute to the general diffused illumi- 
nation in the room. The useful illumination at a point, therefore, 
is materially greater than that obtained only from the nearby 
lights, but materially less than the theoretically integrated effects 
of all the lights. From a practical standpoint, the presence of 
these latter luminous sources is equivalent to raising the diffuse 
component of the light to a point somewhat greater than that 
which would be indicated merely by the coefficient of reflection 
of the surfaces applied to the lights really effective in direct 

Experience shows that in dealing with rooms of this class under 
ordinary conditions of lightness of wall, one must supply light on 
the basis of 1.5 to 2 square feet per rated candle power. In other 
words, the room considered would require from 900 to 1200 candle 
power carefully installed under conditions of the best efficiency in 
order to reach the requisite degree of illumination, a little more if 
illumination for close clerical work is the chief object, a little less 
if the few points where especially good illumination is required are 
treated by the installation of desk lights. Ordinarily it is well to 
get along without individual lights upon a desk if possible, since 
in a large room, particularly if used by a considerable number of 
people, desk lights are a nuisance both from the multiplicity of 
cords required and from the fact that some of them are sure to be 
shining into the eyes of those not immediately using them. 

The arrangement of the outlets must depend on the size of the 
units chosen, and somewhat on the arrangement of the working 
space beneath. The ground plan of the room in question is shown 
in Fig. 134 and the requirements generally would be well met by 
installing, say, 18 60-watt tungsten lamps 7 to 8 feet above the 
working plane, equally spaced on 10-foot centers. These lights 
should be in translucent reflectors of moderate angle, with the 
lower edges carried down far enough to prevent direct light from 
the unscreened filament from shining in the eyes of the workers. 
Most commercial reflectors fail in this particular, the lower edges 
being just too high for adequate protection. The writer has often 




found it good policy, for example, in using the holophane reflectors, 
to install each lamp in the shade designed for the next larger size 
of lamp in order to get the requisite depth. The arrangement of 
lights here suggested is shown on Fig. 134 in the left-hand half 
of the figure. If, for instance, a wide center aisle is in the room, 
six 100-watt lamps would well answer the requirement, or for 
extreme cases six 150-watt lamps, as shown on the right-hand half 
of the figure. 

This type of lighting is very commonly and successfully used 
in large offices and in shops. In the former case one has to look 
out rather carefully for the position of the working spaces to be 
illuminated. In a counting room, for example, it is generally 

























Fig. 134. 

found that after the installation is complete there follow com- 
plaints, from some few desks, of insufficient light.' When examined 
these are almost always due, not to insufficient light, but to 
wrongly directed light, some of the desks being in positions giving 
strong head or hand shadows, which prove annoying. A few very 
simple changes in the desk positions will nearly always remedy 
the trouble. 

The case is an important one because outlets are commonly 
fixed before the use to which the room is to be put is determined. 
In designing outlets it is better to err on the side of too great 
rather than too small a number. Now and then conditions are 
such that some local lighting must be installed, but generally only 
to a very limited extent. 


In the few instances where local lights are to be use4 to a con- 
siderable extent, it is best to make preparations for them by having 
an ample number of baseboard and floor plugs, and then to arrange 
the overhead outlets so as to provide a general illumination of per- 
haps 1.5 foot-candles to reenforce the desk lights. These latter 
should not be overdone. An 8-c.p. carbon lamp, or the smallest 
available tungsten or tantalum lamp, installed under a 7-inch or 
10-inch green-flashed porcelain shade, is ample; and great care 
should be taken to avert the use of larger lamps, which are quite 
likely to result in eye trouble for the users from the glare on the 
paper if not from lights misplaced so that they shine into the 

At this point it is pertinent to inquire concerning the use of 
indirect lighting in such rooms as those here under consideration. 
As already explained, indirect lighting can be carried out by two 
methods, — by concealing the light in coves or similar locations, 
or by installing the lamps in special fixtures adapted to throw the 
light on the ceiling for redistribution. The efficiency of the two 
methods is practically about the same, provided there is equal care 
in design. In either case the indirect lighting is much less efficient 
than the direct, as is necessarily to be expected from the circum- 
stances in the case. 

Assuming fairly light-colored ceilings and walls, experience shows 
that the light required with either indirect system for a given illu- 
mination on a working plane is nearly double that demanded for 
the same illumination from lamps lighting the space directly and 
equipped with suitable and efficient reflectors. Marks (Baltimore 
Lectures, Vol. II, page 702) indicates practically these figures, and 
the author's own experience confirms them. In one recent ex- 
perience in the author's practice both methods were actually tried 
and the illumination and energy carefully measured. The direct 
installation was carried out with reflectors which were deep enough 
fully to protect the eyes. The ceiling and walls were fairly light 
in tone and the space was approximately 2500 square feet. With 
the direct system of illumination, the rated candle power of the 
lamps amounted to 0.24 candle power per square foot per foot- 
candle. With the indirect system under exactly the same condi- 
tions, the light required was 0.45 candle power per square foot per 
foot-candle. In general terms,. with the indirect lighting nearly 
1.8 times the candle power was required for the same effective illu- 


mination. Cravath in a similar test in a smaller room recently 
obtained 1.76 for the equivalent ratio. Bearing in mind the fact 
that both coves and inverted lighting fixtures suffer much from 
dirt, it is well within bounds to say that under average conditions 
the indirect system requires about twice as much light as a well- 
planned direct system for equal illumination. The usefulness of 
indirect lighting is, therefore, rather special than general. When 
properly applied it is pleasant and effective, but never economical 
as compared with well-arranged direct lighting. 

One may next profitably take up the equipment of such a room 
as shown in Fig. 134 when used for manufacturing purposes. The 
old method of lighting a space occupied by machines and workmen 
was to put an individual light shining directly on each machine or 
such part of it as needed special illumination, leaving the rest of the 
room in darkness. The result is what has come to be known as 
" spotted " lighting and the general results are usually bad. The 
strain on the eyes of the workmen owing to the constant transition 
from bright light to darkness as the eye shifts from the work to the 
space beyond is exceedingly trying, much in the same way that 
sudden and violent flickering of light is trying. More than this, 
there are very often intensely bright reflections from parts of the 
machine or work, which dazzle the eye, cause the pupil to contract, 
and interfere very seriously with efficient vision. It has not infre- 
quently been found that workmen complain of insufficient light 
under such circumstances when the actual intensity on the work 
is two or three times that needed on the most liberal estimate for 
proper vision. In fact, the more light the more dazzling the effect, 
and the less effective the vision obtained. 

The remedy for spotted lighting is diffusion, but in buildings used 
for manufacturing purposes a general illumination such as indi- 
cated in Fig. 134 is not often easy to apply. Considerations of 
economy make it undesirable to light at maximum brilliancy cer- 
tain parts of the room, since these parts may be used only for 
storage, for passageways, or for rough work requiring no strong 
illumination. The problem, therefore, reduces itself to lighting all 
parts of the room efficiently for their uses, and this frequently 
implies lighting special machines with considerable brilliancy while 
leaving the general illumination low. The secret of success is so 
to distribute the light as to leave no dark corners and no dark 
regions upon which the eye has to fall, while yet providing at the 


points needed sufficient light for the most critical work that has 
to be done. 

This can best be accomplished by using local lighting at the 
machine plus a general illumination to relieve the contrast of 
darkness and light and to furnish for general purposes a groundwork 
lighting. This end may be reached by the installation of lights 
with translucent shades which permit part of the light to pass 
through for general use, while the rest is directed upon special 
operations which require light. The only difficulty with this pro- 
cedure, which may be carried out by lights with shades of opal 
glass or of prismatic glass with diffusing surface, is that translucent 
reflectors are fragile and in some cases breakage would be a serious 
item. The same result can be attained by using comparatively 
small lamps in suitable reflectors placed near the work and adding 
general illumination from lights equally spaced over the ceiling 
on cords or short stems. The general lighting obviates most of 
the spotted effect which is so unpleasant, but equally essential 
is the reduction of the extreme intensity produced by local lights 
under ordinary conditions. The space shown in Fig. 134, when 
used for ordinary manufacturing purposes, will require on the 
whole about the same amount of light as already indicated, but 
about one-third of that light should be devoted to general illu- 
mination, the remainder being distributed over the machines. 

Lighting machines is a task standing somewhat by itself, because 
while the total amount of light is generally moderate, it must be 
directed where it will do the most good, and machines often have 
projecting arms or other parts which throw strong shadows and 
interfere with the proper observation of the work. Sometimes 
conditions can best be met by small lamps, usually not over 8 candle 
power, with reflectors directing the light where it will do the most 
good. In other cases the lights can best be placed directly over the 
machines, or sometimes with diffusing screens behind the machines, 
the main point being that, while from 3 or 4 up to 8 or 10 foot- 
candles may at times be necessary for the work, the light should 
not be placed so as to throw the work into shadow or to cast 
bright reflections into the eyes of the workman. In some in- 
stances indirect lighting can be resorted to, utilizing the light ceiling 
and walls to the utmost. It should be used, however, only when 
a reduction of shadow is desired with but a moderate degree 
of illumination. The long and short of it is that manufacturing 


spaces must be treated symptomatically, bearing in mind the 
necessity of enough generally diffused light to prevent the spotted 

Occasionally one encounters intricate machines extremely diffi- 
cult to light, and one then has to resort to unusual means. A 
recent case in the writer's experience was the illumination of the 
drying rolls of a paper mill, which after considerable experiment 
was successfully carried out by seeing to it that the interior of 
the hoods over the rolls was painted a good white and then hang- 
ing mercury arcs within the hood, the mechanism being left outside 
and the tubes themselves being suspended 3 or 4 feet below the 
top of the hood with a free opportunity to radiate light in every 
direction. Such a combination of direct and indirect light cuts 
off all sharp shadows and allows one to see in and about the parts 
of a complex machine very easily. These, however, are special 
cases. In the majority of instances a general illumination of 1 foot- 
candle or thereabouts plus 3 or 4 foot-candles in the vicinity of the 
work gives good results in industrial lighting. 


Concentrating reflector 73 

Concentrating prismatic 76 

Concentrating mirror 88 

Diffusing reflector (dense) 44 

Diffusing reflector 63 

Diffusing reflector 63 

Diffusing reflector 48 

Diffusing reflector .- 50 

Diffusing reflector 53 

Diffusing balls 35 

Diffusing balls 36 

Diffusing balls 34 

Half-globes, prismatic 51 

Coves, indirect 36 

Coves, indirect 15 

Indirect 28 

Indirect 29 

Indirect 35 

Semi-indirect 38 

Semi-indirect (inclosed arcs and diffusers) 49 

Arc with opal globe 45 

As a guide in designing the illumination of rooms, large or small, 
presenting no extreme features in the height, shape, or finish, a 
convenient figure to remember is that with lamps arranged at 
ordinary heights near the ceiling and equipped with well-designed 
diffusing reflectors, one can count on receiving upon the working 


plane 4 or 5 foot-candles per rated candle-power-per-square-foot. 
This implies that about 50 to 60 per cent of the total flux of light 
is utilized on the working plane, the rest being taken up by ab- 
sorption and unutilizable reflection. The table on page 245, derived 
mainly from the writer's own experience, gives the actual efficien- 
cies of utilization reached in lighting installations with various 
types of shades and reflectors. 

These figures will serve as a useful guide in computing the 
illumination by the flux o£ light method. It will be noted that 
the highest efficiencies are with somewhat concentrating prismatic 
or mirror reflectors arranged to throw as much as possible of the 
light flux upon the working plane without any material utilization of 
the diffusing surfaces of the room. The next grade of efficiency is 
obtained with diffusing reflectors of opal or prismatic glass which 
throw a large amount of the light downwards and yet transmit a 
material fraction which is rediffused by the walls. A slightly lower 
grade is given by the semi-indirect system, which rises in efficiency 
as more of the light-flux is sent directly to the working plane 
and less diffused from the ceiling. Of about the same efficiency 
are the lights inclosed in diffusing balls or globes, the thinner globes 
of course being the more efficient. Next in order come the purely 
indirect systems in which none of the light is sent directly to the 
working plane but all is diffused from surfaces which are as a rule 
none of the best. 

: Any one of the systems indicated can be made to give thor- 
oughly good illumination and any one of them can be so misused 
as to be unsatisfactory. As between the ordinary direct and the 
indirect systems of lighting, the former when properly installed 
are always the more efficient. One interesting question which has 
arisen as between the direct and indirect schemes of lighting is 
the quality of the illumination as regards its usefulness. There 
is no physical or physiological reason why there should be any 
difference in the usefulness of a given illumination derived from 
either method of arrangement, assuming each to be planned with 
equal skill in the way of avoiding, on the one hand, glare from 
exposed sources, and, on the other, a perhaps equally trouble- 
some glare from over-illuminated surfaces. There have been some 
strong expressions of opinion by those who have had a prejudice 
in favor of one or the other method. The first figures derived 
from actual experiment on the matter are those of Millar ("Trans- 


actions Illuminating Engineering Society/' Vol. II, page 583), 
which indicate that a given intensity of illumination derived from 
indirect lighting is materially less effective than when derived 
from direct lighting. On the other hand, some recent observations 
by Cravath indicate exactly the reverse, based on judgments made 
like Millar's by a group of independent observers. The psycho- 
logical and casual physiological factors in the case are, however, 
so uncertain and variable that it is unsafe to generalize from either 
of these diverging statements. In the author's opinion, whatever 
differences have been observed are due to secondary rather than 
primary causes, and must disappear when the installations are 
really skillfully planned. The difference in efficiency of utiliza- 
tion is, however, unavoidable. 

Where merely rough work is being done, arcs may be effectively 
used, always, however, shaded by ground or similar globes. These 
are distinctly cheaper, because more efficient, than carbon incan- 
descents, but their light lacks the steadiness desirable for work 
requiring close attention. Six 350-watt arcs would give, in the 
room shown in Fig. 133, very good illumination, when placed in 
approximately the positions deduced for the six clusters, with a 
total expenditure of 2100 watts as against about 4500 watts 
required by the clustered incandescents, and, say, 3600 watts 
required by about 36 pendent 32-c.p. lamps. In many cases, 
less light than this would be required, and the total amount of 
energy could be correspondingly reduced. As already indicated 
tungstens would do even better. 

From Fig. 133 it appears that in using arcs about 2000 to 2500 
square feet may be assigned to each 500-watt arc, and 1000 to 
1500 square feet to each 350-watt arc. It should be remembered 
that the inclosed arcs with inner globes are nearly 25 per cent 
less efficient than this, although to be preferred by reason of their 
ordinarily greater steadiness, and that alternating arcs are slightly 
less efficient than continuous-current arcs. 

Arcs do their best work when placed fairly high and used in 
cases where protracted close attention on the part of the workmen 
is not necessary. They are preferable to incandescents of any 
kind when colored objects are to be illuminated. 

In workshops where special objects are to be illuminated, arcs 
are at a great disadvantage with respect to the distribution of 
light, since their relatively small number forbids placing them 


in the most advantageous positions with respect to all the 
machines. f 

They have, in short, the disadvantage of being radiants too 
powerful for the best distribution. It is thus found that in 
practical illumination arcs are considerably less efficient than 
their actual candle power would indicate. The effect of the 
bright radiant upon the eyes, the rather dense shadows, and 
the slanting light at a distance from the arc, unite to produce 
results that cannot be predicated from photometric measurements 

For example, a 350-watt open arc is, in point of mean spherical 
candle power, closely equivalent to ten 32-c.p. incandescent lamps; 
but in an actual installation indoors there are few cases in which 
the arc could not be replaced . by six such incandescents without 
detriment to the illumination. With tungsten incandescent lamps 
at 1.25 watts per rated candle power, the ordinary carbon arc 
compares unfavorably except in cases where its whiter color is 
important. For such cases the intensified arc should generally 
be used. 

If strong illumination is the object to be attained, there is 
little doubt that for gas lighting in rooms of the size considered, 
mantle burners should invariably be used. As already intimated, 
each such burner of the ordinary size is equivalent to about two 
16-c.p. distributed incandescents. If the lamps are grouped in 
each case, the mantle burner must be given a rather better rating, 
being equivalent to between 2.5 and 3 such incandescents. Prop- 
erly shaded, the mantle burner is a very economical and effective 

For lighting large areas, like those we have been considering, 
it is very well adapted, but if the lights are placed high it is neces- 
sary not only so to shade them as to hide the mantles, but they 
must, in addition, be furnished with such shades or reflectors as 
will throw the light effectively downward. Reflectors or holo- 
phane globes used with the mantle burners will correct this faulty 
distribution and enable them to be used more effectively in the 
case in hand. The modern inverted Welsbachs diminish this diffi- 
culty and give excellent results, but even with these it is neces- 
sary to use diffusing reflectors to shade the eye and improve the 

With higher rooms than usual, one can concentrate the radi- 



ants more advantageously, and has considerably more liberty of 
action in placing the lights. 

Fig. 135 is intended to illustrate the conditions which exist 
in a very high room of fairly large area. It shows in vertical 
section a room supposed to be 50 feet square and 50 feet high, 
the plane of illumination, ab, being 3 feet from the floor. We 
have here 2500 square feet of floor surface. At the ordinary rate 
of 3 square feet per candle, this would demand 833 candle power, 

Fig. 135. — Vertical Section of Hall. 

or practically 52 16-c.p. lamps, or, with a coefficient of diffusion of 
1.50, about 36 such lamps. 

But the previous calculations having been made for a room 
only one-half this height, and with lamps placed considerably 
below the ceiling, it is clear that the greatly increased height in 
the present case will lead to somewhat different conditions unless 
the lamps are to be dropped very far below the ceiling — so low 
as to produce a decidedly unpleasing effect. Lamps placed, for 
example, in the plane cd, corresponding to frieze lamps in the 


previous instance, are too low to look well, while they would, on 
the basis just given, furnish the room with satisfactory illumi- 
nation. If placed on side brackets at or below the plane cd f they 
would work well on the floor, but would produce the effect of the 
ceiling fading into dimness unless the ceiling itself had an extremely 
light finish. 

Such a room, therefore, while very easy to light thoroughly, is 
very difficult to light both thoroughly and with good artistic 
results. Rooms of such dimensions are seldom used for manu- 
facturing purposes, these shapes occurring "more frequently in 
rooms for public uses of various kinds. 

Without going into detailed computation, which the reader can 
readily make for himself, in the light of previous work on Fig. 
134, it is safe to say that often the best general effects would 
be produced by placing perhaps one-third of the total candle 
power in lamps of moderate candle power, as a frieze or in clus- 
ters, 8 or 10 feet below the ceiling, in the line ef, or thereabouts, 
and putting the remainder on brackets, in groups of three to six, 
a little below the plane cd. Such an arrangement obviously loses 
somewhat in the efficient disposition of light, on account of the 
great height of part of the lamps, which can be depended on only 
for a rather faint groundwork of illumination on the plane of 
illumination rib. If, for example, the total installation consists of 
600 candle power, of which 200 is in the frieze, the mean distance 
of the frieze lamps from a point, say, in the middle of the floor, 
would be in the vicinity of 45 feet. 

Consequently, allowing for the effect of the reflectors of the frieze 
lamps, and for what each can do by diffusion, it is safe to say that 
the frieze lamps would give an illumination of not over one-fifth 
foot-candle on the plane of illumination. Hence, something like 
eight-tenths foot-candle would have to be furnished by the lights 
upon brackets. The amount of light furnished by these would, 
therefore, have to be about eight-tenths of the total illumination, as 
determined by lights placed in the relative position shown; that is, 
the ceiling lights of one-third the total candle power really would 
be furnishing not over one-fifth of the total light, which means 
that for lights placed as just indicated the total candle power in- 
stalled should be increased somewhere from 25 to 33 per cent, or 
rather more, as the bracket lights cannot always be conveniently 
placed in favorable situations. 


Hence, in a room so illuminated, it would not be safe to allow 
more than 2 to 2.5 square feet of floor space per candle power, and 
generally nearer the former figure than the latter. To attempt the 
lighting of such a room by frieze or ceiling lights, as ordinarily 
placed, would be wasteful. If economy is not an important factor 
in designing the illumination, at least half the lights might be placed 
in the frieze with a distinct gain in artistic effect. In such case 
the total installation should be fully 50 per cent greater than the 
minimum required. We shall see, however, that there are effective 
methods of getting a strong groundwork illumination from above 
without resorting to either of these methods. 

To follow up the effect of raising the lights in a high room still 
further, it is well to note that the critical point is the amount of 
available diffusion. If one were dealing with a room lined with 
black velvet, or with translucent walls, in which there is only a 
very minute amount of diffused light, raising the lights would 
diminish the illumination quite nearly according to the law of inverse 
squares, assuming unshaded or similarly shaded lamps. 

Writing now K for the coefficient of diffusion denoted by the 

fraction f _ , J, and recurring to the formulae previously given for 

illumination, we have at once KC =IxP, and for fixed values of 
C and L, d = P Va , where P is a constant. Hence we may con- 
clude that for any desired value of the illumination with a fixed 
amount of lights available, the height to which these lights can be 
raised and still produce the required effect is approximately pro- 
portional to the square root of the coefficient of diffusion. 

The moral of this is tolerably obvious. If one deals with a dome 
finished, let us say, in white and gold, it may be permissible to 
place a large part of the lights fairly high up, while in a church 
with a vaulted roof in dark oak and with dark walls, lights placed 
high are nearly useless for purposes of illumination. In such a 
case, lights placed at the level of the roof beams and unprovided 
with reflectors have barely more than a decorative value, and 
should be treated, if used at all, essentially as a decorative feature, 
useful for bringing out the details of the architectural design. 

Any real illumination must be accomplished by lamps with reflec- 
tors or by lamps placed down nearer the plane of illumination. 
In these dark interiors lamps with reflecting shades can be used to 
especial advantage, since the coefficient of diffusion is so small that 



the lessened diffusion due to the partially directed beams from 
reflectors is of trivial consequence. In fact, there are few cases in 
which reflectors cannot be used to advantage in rooms having very 
high ceilings. 

Churches are generally badly lighted, and are, in fact, rather 
difficult of treatment, if of any considerable size. They are seldom 
brilliant in interior finish, usually have rather high vaulted roofs, 





^ eu\ r 







-40- > 

Fig. 136. 

and require fairly good reading illumination. The few cases in 
which their form approximates to Fig. 134 may easily be treated 
as there indicated, but such is not the usual condition. Fig. 136 
gives a roughly typical church floor plan as regards the main body 
of the building. The total floor space is shown as 5000 square feet 
in the nave and choir combined, and 800 square feet in each tran- 
sept. The walls are assumed to be 30 feet high in the clear, with 
a Gothic roof above. Now the total area to be lighted is 6600 


square feet, and the value of K is low, not safely to be taken as 
exceeding 1.20. The peculiarities of the building, as a problem in 
lighting, lie in the high walls and the absence of any ceiling, both 
of which complicate matters. 

As to the nature of the radiants, when electric lights are avail- 
able one must depend almost entirely upon incandescents. Arc 
lamps are not to be considered for artistic reasons, save, perhaps, 
in indirect lighting of the choir. If only gas is available, mantle 
burners suitably and thoroughly shaded had better be the main 
reliance, as ordinary gas flames are seldom steady in such a place. 
In either case it is generally wise to avoid chandeliers. The only 

Fig. 137. 

form of chandelier for which there is good historic reason in Gothic 
churches is the great couronne, like those found at Hildesheim and 
elsewhere, originally symbolic of the Heavenly City, and exceedingly 
ornate, having, therefore, the excuse of an intentional decorative 
and ecclesiastical value. Fig. 137 shows the Hildesheim couronne, 
which has recently been fitted with incandescent lamps. 

As to the amount of light needed, it would be advisable to 
allow no more than 2.5 square feet per candle power, which, taking 
A' at 1.20, would call for 2200 net candle power. In point of fact, 
in using electricity, not less than 150 16-c.p. lamps or their equiv- 
alent should be used, and even this number, on account of the 
trying conditions, would have to be very deftly arranged to give 
the required result. For the best effect they should be chiefly 


lamps with diffusing globes, assigned about as follows: 90 to the 
nave, 20 to the choir, and 20 to each transept. As to position, 
the most efficient method would be to put them in groups of six 
or eight on brackets between the windows, at half to two-thirds 
the height of the wall, with possibly larger groups massed at the 
four corners of the crossing. With still more lights available, very 
beautiful results could be attained by adding lights at the capitals, 
and, in some cases, along the tiebeams, or on the corbels from 
which the pendent posts rise. These latter arrangements are very 
effective, but not economical, and if used should be installed pn 
the basis of about 1 candle power per 2 square feet of floor surface. 
With the large tungsten units the number may be proportionally 
smaller, which simplifies the fixture design. All incandescent 
lamps used without diffusing shades should have ground bulbs. 

The chief point in church lighting is to furnish modest reading 
illumination, say 1 to 1.5 foot-candles, without glaring sources in 
the eyes of the congregation and without breach of the archi- 
tectural unities of the place. The treatment must be almost 
entirely guided by the individual situation, and is often hampered 
by the existence of fixtures or particular developments of methods 
which cannot well be gotten rid of. Now and then standards, 
bearing clusters of lamps, are installed throughout the nave of a 
church, and occasionally these standards are good enough to make 
their retention desirable. By fitting them with small incandes- 
cent lamps within small diffusing globes very pleasing results can 
sometimes be reached. As a rule, lights should not be carried 
high in churches. The mediaeval church, closely copied in many 
modern interiors, was a place where little light was necessary, 
and that little was needed in the lower body of the church, the 
towering roof gaining rather than losing in effect by fading into 
obscurity above. Then, too, most worshipers either were unable 
to read, so that they needed no reading illumination, or knew the 
service, so far as they were required to know it, by heart, so that 
again light was unnecessary. 

In modern churches and forms of worship these conditions have 
so changed that a good reading illumination is necessary, as has 
been before remarked, but it is not generally furnished. In fact 
church fixtures, on the whole, are the least adapted to their use 
of any that can be found even by patient searching. The mischief 
is generally done by stupid and slavish following of inappropriate 


precedents. The lighting of the Mosque of St. Sophia in Con- 
stantinople has probably been responsible for more badly illumi- 
nated churches than any other one malign influence. The lighting 
fixtures in this famous building are shown in their general bearing 
in Fig. 138. They consist of spreading wrought-iron and bronze 
fixtures carrying numerous tiny oil lamps, not far over the heads 

Fig. 138. 

of the worshipers — about 9 feet from the floor. Each fixture is 
borne by a preposterously long rod hanging from the huge and 
lofty central dome, a method of support made necessary by the 
extraordinary proportions of the building and the conditions of 
its use. This kind of fixture has been in use there for centuries. 
But whatever venerable association may consecrate it now in the 
eyes of Christian and Mohammedan alike, there is no possible 


propriety in copying so clumsy a device in modern churches of 
different architectural character, with all our present illuminating 
resources at hand. 

In the hands of the Occidental barbarian, this type of fixture 
usually degenerates into a short bronze or iron barrel, within which 
is ill concealed a group of glaring incandescent lamps throwing 
their light into the eyes of the just and unjust alike, and giving 
a maximum of glare with a minimum of illumination. The only 
possible method of reducing the glare is so to inclose the lights 
as to give almost no illumination, and this has also sometimes 
been done, less through design than through sheer stupidity. 
The writer calls to mind at least a dozen churches in which such 
fixtures are conspicuously useless, and, he may add, has had the 
pleasure in several instances of throwing them out and replacing 
them by devices less offensively glaring and far more effective. 

The nav$ and the transepts of a church may be easily lighted 
in accordance with the methods earlier suggested, preference being 
given to that which lends itself most harmoniously to the archi- 
tectural requirements. Above the side aisles it is often desirable 
to use lights with small diffusing globes, placed close to the ceiling, 
although sometimes the height is sufficient to enable brackets to 
be used, if also employed for the main work of lighting the nave 
and transepts. 

The sanctuary is a different matter. In churches having an 
elaborate ritual bright lighting here is desirable, especially in 
connection with the altar. Generally there is an opportunity for 
placing lights behind the chancel arch, either above, or at the sides, 
or in both situations. Where the available space permits, such 
lamps can with great advantage be furnished with somewhat con- 
centrating reflectors arranged to flood the sanctuary with light 
when desirable, and particularly so to illuminate the altar as to 
bring out its full decorative value. Sometimes it is desirable 
to specialize the lighting for these two functions, so that the altar 
can on occasion be thrown into striking prominence. It must 
not be forgotten that most elaborate altars should not be lighted 
uniformly from the front, inasmuch as this tends to suppress the 
detail, which is often their chief charm. To avoid this difficulty, 
lights with focusing reflectors can be massed on one side of the 
space behind the chancel arch so as to illuminate the altar at any 
required angle, leaving enough general illumination to prevent too 


dense shadows. The particular arrangement of course depends on 
the special things which it is desired to bring into prominence. 

The lights in the sanctuary should be upon several switches 
so arranged as to secure any requisite intensity of illumination 
for the various cases which have to be met. Where candles are 
freely used in connection with the ritual, the question of replacing 
them by incandescent lamps frequently comes up. Decorative 
standards used around the altar can often be fitted for small-bulb, 
heavily frosted incandescents with good effect, but where the can- 
dle has a symbolic value in the ritual such a substitution is in very 
dubious taste. One recoils from the thought of the Tenebrae with 
the candles turned off by snap switches on the wall. 

Suitable illumination for the reading desks goes almost without 
saying. It should be given by carefully shielded reading lights 
serving the purpose of their use without attracting any further 
attention. Non-ritualistic churches having the form of a large 
hall require the lighting appropriate to that case, and do not 
involve any of the special problems inherent in other churches. 
The question of stained-glass windows is one which sometimes 
arises as a problem in illumination. Comparatively little has been 
done in this matter save in isolated cases, but there is no reason 
why, if thought desirable, the chief windows may not be illu- 
minated from the outside by indirect lighting. In this case arcs 
are by far the most convenient source of light, since they alone of 
common illuminants give a light Sufficiently rich in blue to bring 
the stained glass to its daylight value. In cases where there is 
strong daylight lighting from a dome, it may be desirable by 
indirect illumination to bring the interior at night to something 
approaching the condition best suited to displaying it by day. 
Aside from such uses as these, the arc light has no place in a 
church, although even flame arcs have been used, by a combination 
of bad taste and bad judgment, for church lighting. 

In lighting with gas, brackets are about the only thing feasible, 
since the flames must point upward, and few capitals would fail to 
look overloaded with adequately shaded burners. Mantle burners, 
of course, do the work most efficiently, but used alone the effect is 
certain to be grimly utilitarian; and especially around the choir 
small ordinary jets may be used to very great advantage. The 
mantle burners should be as unobtrusive as possible in such a case, 
even if they do the main work of the illumination. 


Only the barest hints can be given for the detail of church light- 
ing, as so much depends on the architectural peculiarities and on 
the scheme of decoration, but the foregoing indicates the general 
principles to be followed. The most important thing is to give a 
rather moderate illumination without the individual radiants ob- 
truding themselves unpleasantly on the eyes of the congregation. 

Large public buildings are generally easier to light than churches, 
since they are, as regards the shape of the several rooms, com- 
paratively simple and are seldom dark in finish. Many rooms may 
be illuminated along the lines already laid down, but, on the whole, 
powerful radiants, such as arc lights and the largest tungsten 
lamps, may be more freely used here than elsewhere. 

In very high corridors and high halls without galleries arc lights 
can be used with excellent results. They should invariably be 
shielded by ground or opal globea, and, if hung very high, as is 
generally desirable, to keep them out of the ordinary field of vision, 
should be provided with reflectors. They should be numerous 
enough to suppress the shadows that ordinarily exist under the 
lamps. From the absence of such shadows the modern intensified 
arcs have a very material advantage. 

Rooms lighted by arc lamps ought to be of light finish, since 
the lamps must be placed rather high to keep them, even shaded, 
from glaring unpleasantly; and they give a strong, nearly horizontal 
beam, which, in lack of good diffusing surfaces, is for the most part 
wasted. Reflectors deep enough to turn this downward would 
usually be most unsightly and would give an unpleasant search- 
light effect, which should be avoided. 

Never let the eye rest simultaneously on arc and incandes- 
cent lamps indoors, since the latter seem very dim and yellowish 
in such company, and will never be credited with anything like 
their real brilliancy. Similar reasoning applies to the use of mantle 
•burners and ordinary gas jets in the same room. When so used 
the former should be well shaded and unobtrusively placed, and 
the latter massed and generally unshaded or lightly shaded, so as 
not to seem of relatively very small intrinsic brilliancy. 

Sometimes in large interiors the powerful regenerative burners 
and high-pressure mantle burners may find a place. They give 
an excellent downward illumination, which is occasionally very 

Theaters present some very interesting problems in illumination 



on account of their peculiar shape and the difficulty of lighting 
the interior with sufficient brilliancy without making the radiants 
altogether too conspicuous. They are, as a rule, more brightly 
lighted than other interiors, but seldom judiciously. The usual 
fault is to place the lights so that they shine directly in the eyes 
of a considerable part of the audience. The auditorium is com- 
monly very high in proportion to its area, and plentifully supplied 

■ ■ 

fig. 139. — Elevation of Theater. 

with galleries. Fig. 139 shows the typical elevation, the floor plan 
being generally only slightly oblong. The galleries, of course, 
sweep around the sides, narrowing as they near the proscenium 
boxes. Not infrequently a fourth gallery is added. 

During the acts no very considerable amount of light is needed, 
but between them it is generally desirable to produce an effect 
of great brilliancy. The main floor is far below the roof, and 
the shelving galleries render it difficult to light the spaces be- 
tween them. The general fittings are usually light, but the dull 


hue of the floor and galleries when occupied kills much of the 

The actual floor space to be dealt with as a problem in illumina- 
tion includes the galleries, and hence greatly exceeds the area of 
the main floor. Assuming the width in Fig. 139 to be 50 feet, 
the nominal area in front of the footlights is 3000 square feet. 
The total gallery area is usually from 1 to 1.5 times the floor space, 
so that the entire space to be lighted would be at least 6000 square 
feet, half of it being located so that it can get little advantage 
from the illumination of the main space above the floor. The 
space behind A, and the galleries B, C, and to a less extent D, have 
to be treated almost as separate rooms, particularly when, as some- 
times happens, the galleries are rather lower than shown in Fig. 139. 

This is the main reason for the apparently abnormal amount of 
light that is needed in theaters. The fact is that there is really a 
very great area to light, and it is so placed that it cannot readily 
be treated as a whole. The following table shows the approxi- 
mate amount of illumination furnished in a number of prominent 
Continental theaters. 

If in Fig. 139 we allow, on account of the high ceiling and con- 
ditions unfavorable for diffusion, 2 to 2.5 square feet per candle . 
power, and take account of the real total floor space, including the 
galleries, we reach just about the figures given below, which are 
based on the floor plan only. And in practice 3600 candle power 
would probably do the work well, although, since this only allows 
ordinary good reading illumination, more light is necessary to give 
the really brilliant effect which is usually desired. Fully 5000 
candle power would be required to show off the house effectively. 

Theater. Sq. Ft. per C.P. C.P. per Sq. Ft. 

Opera, Paris 0.78 1.28 

Opera, Paris, as ballroom 0.38 2.63 

Odeon, Paris 1 .52 0.66 

Gaiety Paris 1.14 0.87 

Palais-Royal, Paris 0.51 1 .96 

Renaissance, Paris 0.52 1 .92 

La Scala, Milan 1.07 0.93 

Massimo, Palermo (ordinary) 0.86 1.16 

Massimo, Palermo (en tete) 0.53 1.88 

As to the location of the lights and their character, the body 
of the house can be usefully lighted by lamps ranged along the 
galleries at abc. If these are placed below the edges of the galleries, 
they will glare directly into the eyes of the spectators, so that it is 


better to illuminate the gallery spaces from the rear and above, at 
a'b'c'. The radiants may well be provided with reflectors, as the 
diffusion amounts to little, and all lamps on and under the galleries 
should have ground globes. These lights may be reenforced to 
great advantage by ceiling reflector lamps, best sunk in the ceil- 
ing deep enough to make them inoffensive from the galleries. 
These, with some ornamental lighting about the stage and boxes, 
should give a capital result. The main point is to light the inte- 
rior brightly without thrusting bright radiants into the field of 

A useful form of ceiling lighting, applicable to many very high 
interiors, is arranged by replacing the lamps at d, Fig. 139, by 
opal-glass skylights of rather large dimensions, and placing above 
them arc lamps with reflectors. The skylight surfaces should be 
flat or slightly projecting rather than recessed, and the reflectors 
should be planned so that each may throw a cone of light sub- 
tending an angle equivalent to the whole floor plan. 

By thus superposing the indirect illumination from a group of 
lamps the general steadiness of the light is greatly increased. In 
thus using arcs, care should be taken to have the diffusing sky- 
lights faintly tinted so as to lessen the color contrast between 
the powerful ceiling lights and the incandescents used elsewhere 
in the house. It is a considerable advantage thus to place lights 
above the ceiling, as it avoids the serious heating effect due to 
massing incandescents near the ceiling of a generally overheated 

On account of this heating the use of gas in theaters is highly 
undesirable, and has been almost completely abandoned. In lack 
of anything better, fair results could be reached by mantle burners 
placed somewhat as shown in Fig. 139, and very thoroughly 
shaded by holophane or other diffusing globes, much of the illu- 
mination being located above the ceiling. 

The Lighting of Schoolhoicses. — The lighting of schoolhouses 
stands somewhat apart from the ordinary illumination of large 
interiors on account of the exceptional uniformity desirable and 
the severe requirements of lighting adequately a large number of 
working spaces arranged in an entirely formal way. First-class 
illumination must be provided for usually about forty pupils, and 
it must be so arranged that there is no trouble from shadows 
of head or hand upon the work, while each pupil must get this 


adequate and well-directed lighting when sitting in a comfortable 

In rooms used for general clerical purposes, as has been indi- 
cated, the usual difficulties may be eliminated by slight changes 
in the positions of the desks or by turning them around. This 
is not permissible in a schoolroom, where the desks are fixed in 
an orderly manner. Moreover, the light must be very carefully 
kept out of the children's eyes, and glare, either direct or re- 
flected, must be carefully avoided. As to the amount of illumi- 
nation required, it must be sufficient for very easy reading and 
writing, the amount depending somewhat on the class of work to 
be done and the sort of books used. A minimum of 2 or 3 foot- 
candles is imperative. For the ordinary class exercises the former 
amount as an irreducible minimum serves well, since the average 
is likely to run well over this figure. 

In case of work with textbooks of more than usually fine print 
or with Greek or Gothic type, this minimum should certainly be 
raised, and the author is inclined to agree with Dr. Broca that 
3 to 4 foot-candles is a better minimum figure, especially if critical 
work is to be done and the hours of artificial lighting are relatively 
long, as, for instance, in the case of rooms used for evening schools. 
The cost of such an increase is trivial, and with well-distributed 
and well-diffused illumination it is far safer to err on the side of 
a high minimum. In some special cases where manual work of a 
somewhat trying character is being done, not less than 5 foot- 
candles is desirable. 

In 1907 the Boston School Committee appointed a committee 
of three oculists and two electricians to examine into the con- 
dition of school illumination and report to the committee. The 
author had the honor to serve this commission in a consulting 
capacity and took part in the experiments tried. 

The standard Boston schoolroom is about 26 by 30 feet, with 
a 13-foot ceiling, and usually contains 42 individual desks besides 
the teacher's desk on a raised platform. Daylight illumination is 
obtained from windows on the left of the desks as the pupils 
sit. The woodwork is usually a light yellowish color, the walls 
of a faint green or buff, and the ceiling white. The coefficient 
of reflection of the walls, when clean, is usually about 0.45. A 
spare schoolroom was fitted up for experimental purposes and a 
large amount of time was spent in trying the effect of various 



arrangements of lights and various types of fixtures. The final 
outcome is shown in Fig. 140. At each of the designated spots a 
40-watt tungsten lamp was suspended 10 feet 6 inches from the 
floor in the diffusing reflector shown in Fig. 141, a simple chain 

x^± — 

4 i 





~1 , i d ' >\ t -g^Mr- l\* — |W-*H- 84m--»fc— i&i- 













Fig. 140. 






~- 3-3H- 




fixture being used. The reflector was of prismatic glass with a 
diffusing enameled coating, and the lamp, which had a frosted 
tip, was located so far within the shade as to keep the light 
effectively out of the pupils' eyes. 

The unsymmetrical positions chosen for the outlets were found 
to be very efficacious in avoiding head arid hand shadows, which 



were at no point troublesome. The position of the lights is 
shifted slightly forward to avoid the head shadows and slightly 

to the left to avoid the hand 
shadows. The illumination thus 
attained was very uniform, be- 
ing approximately 2.5 foot- 
candles at every desk, which 
amount was found to be fully 
adequate for ordinary school- 
room purposes. Approximately 
this scheme of illumination is 
being installed in all the new 
schoolhouses in the city of Bos- 
ton, and the older ones are being 
changed to accord with them. 
This system superseded a 

Fig. 141. 

semi-indirect system in which six clusters symmetrically placed, 
each of four lamps, were inclosed in shallow opal bowls pointed 
upward to secure diffusion from the ceiling and covered with 
plate glass to keep out dust. This system, although taking 
the same energy as the one later adopted, gave only 1.5 foot- 
candles as against 2.5, showing, therefore, about the usual ratio 
between direct and indirect lighting. With 60-watt tungsten 
lamps instead of 40-watt, the illumination is raised to about 3.75 
foot-candles, which is a somewhat better figure in cases where 
much work by artificial light has to be done. It was deemed 
desirable to meet the trying period at which natural light has 
to be abandoned and artificial light used, by making the passage 
from one to the other complete by drawing the window shades 
as soon as artificial light became at all necessary. 

It is interesting to compare this result with that reached by 
Dr. Harman, the oculist of the London County Council Education 
Department and reported in "The Illuminating Engineer" (Lon- 
don). Fig. 142 shows his arrangement of lights in a typical school- 
room to accommodate forty pupils at twenty double desks. An 
asymmetrical arrangement very similar to that found by the 
Boston Commission is the result of his investigation, either four or 
six units being employed. A special light with opaque reflector is 
placed at the position marked X at the left of the master's desk, 
to be used either for the desk or the blackboard, as required. 



The shades recommended are deep enough to shield the chil- 
dren's eyes and whether for gas or electric lighting the arrange- 
ment shown is a suitable one, although necessarily somewhat less 
uniform in its results than the Boston plan with its larger number 
of units. The lighting found on the desks in the better class of 
English schools is found to be in general terms 2 to 4 foot-candles, 
which appears to be a fully adequate amount. In the London 
Arts and Crafts School, the illumination is carried, as it should 
be, considerably higher, ranging from 4.5 to 8 foot-candles and 
reaching a perhaps unnecessary maximum of 30 foot-candles in 
the wood-carving room, where localized pendent lighting is em- 
ployed. These illustrations show clearly the general requirements 
in schoolroom lighting. 


□ CH CD CD □ 


cd nn nn o n 

Fig. 142. 

The lighting of the blackboards is a matter of special concern, 
and in arranging the lights particular care should be taken to 
see that there are no troublesome reflections from the blackboard 
for any point of view. The most effective precaution is to have 
the blackboard lighting well diffused, to avoid the use of shiny 
enameled surfaces on the blackboards, and to see that they are 
kept clean so as to secure proper contrast when the boards are in 
use. The blackboards of the Boston schools are upon the right- 
hand side of the room and adequately lighted from the lamps 
nearest them. 

Lighting Tennis Courts. — It is only occasionally that a large 
and high interior has to be brought to a very high degree of illu- 
mination. Perhaps the most difficult task of this kind is the 



artificial illumination of courts for lawn and court tennis. In 
these cases, particularly if a fast game is being played, the illumi- 
nation must be both high and uniform, and the sources must be 
kept high to be out of the way of flying balls and to reduce the 
glare in the players' eyes. 

The lawn-tennis court is the easier to light, since the walls do 
not come into play. At least 4 to 5 foot-candles is the illumination 
required, and more is better. The best results which the author 
has reached in his own practice have been with mercury arcs. 
In several instances a dirt court illuminated by 12 of the commer- 
cial tubes, each about 22 inches long, equally spaced in two rows, 
with their lengths crosswise the court, has given very satisfactory 
playing conditions. Fifteen such tubes placed in three rows would 




















Fig. 143. 

push the illumination a little higher and would be more satisfactory 
for a court much in use. The tubes are, of course, shielded by wire 
netting to avoid their being broken by flying balls. Their height 
in these cases is about 25 feet. 

An interesting scheme has recently been successfully tried in 
London for this purpose, in which the interior of the building, 
except for the lines of the court and the top of the net, was made 
dead black to secure greater contrast between the balls and their 
surroundings. In this case, 8 nominal 1500 candle-power high- 
pressure inverted gas lamps were used, with reflectors above them 
and ground-glass screens below. These were arranged as shown 
in Fig. 143. The illumination obtained was 4 to 5 foot-candles and 
the installation served its purpose admirably The foot-candle 
readings obtained are shown on the plan. Yellow flame arcs simi- 


larly placed would give excellent results, but they are somewhat 
more dazzling than the mercury tubes or the press-gas lamps. 

The problem of lighting a court-tennis court is considerably more 
difficult. In one instance the author obtained a good result by 
using 18 double 22-inch mercury lamps, equally spaced in three 
rows over the court. The interior color was, of course, greatly 
changed when the artificial illumination went on, but the con- 
trasts obtained were satisfactory and the installation has given 
good results. Whatever be the illuminant chosen, particular pains 
must be taken to get strong light in the extreme corners of the 
court and on the tambour. 

Hand-ball and squash courts are easily illuminated on about 
the same basis as in lawn-tennis courts, but up to the present a 
racquet court defies all attempts at adequate artificial illumination. 
The balls are so small and the pace so terrific that no ordinary 
amount of light seems to produce any useful result. At the most 
modest estimate at least twice as much light is necessary as in 
court tennis, and it is doubtful whether even this would be 
enough for a really fast game. 

The Lighting of Libraries and Similar Buildings. — Libraries 
and museums and similar public buildings demand a somewhat 
specialized illumination both with respect to intensity and distri- 
bution. In a library the first condition is that there should be 
in all parts of the building where general reading is to be done 
light sufficient for reading, easily and comfortably, at any seat in 
the room, any book which may be in use. Second, there must be 
complete absence of glare either from the luminous sources them- 
selves, or indirectly from the paper of the books. Third, for the 
use of bookshelves or in rooms for general purposes, there must 
be a comfortable general illumination irrespective of the positions 
of the readers' seats. 

The first requirement is a somewhat severe one, since lighting 
which, for example, is quite sufficient for a school desk will be 
found insufficient when reading books in fine type or studying 
critically maps and engravings. Among library authorities who 
have acquainted themselves with illuminating conditions, and 
among illuminating engineers who have worked on this problem, 
there is general agreement that illumination up to nearly 5 foot- 
candles is necessary at the reading tables. For general use about 
the room 1 or at most 2 foot-candles is a liberal allowance. 


The natural inference from these requirements as to intensity 
is that the best system for illuminating the reading rooms of a 
library is a combination of a general illumination with localized 
lights on the reading tables. This is certainly the most economical 
solution of the problem and perhaps the most generally applicable. 
For this purpose well-shaded lights near the ceiling may be utilized 
to produce 1 to 2 foot-candles over the whole working area. On 
the reading tables lamps carrying the sources of local illumination 
should be placed, so shaded as to keep the light out of the readers' 
eyes, and to avoid as far as possible strong direct reflections from 
the paper. 

The first part of the task, that is, general illumination, is a very 
easy and obvious matter. The second is not. In many libraries 
rather wide tables with fixed table lamps along the center line are 
used, and sometimes with extremely bad results. A fixed lamp on 
a wide table if at a sufficient height to spread the light the full 
width of the table is extremely likely to shine directly into the eyes 
of persons below the average height when seated, and to produce 
serious glare by direct reflection of the light from the book. 

In one library with which the author had to deal, the glare pro- 
duced in these ways was so serious as to make reading very un- 
comfortable, and the situation was far from being relieved by the 
use of unshaded, though frosted, incandescent lamps for general 

The remedy applied was to place these lamps in diffusing balls 
and to substitute for the fixed standards movable student lamps 
with deep 10-inch porcelain shades, flashed with green on the out- 
side, and containing 16-c.p. lamps, frosted, well up toward the top 
of the cone. The position of these shades was adjustable, and the 
lamps themselves, being on flexible cord, could be shifted to suit the 
requirements of the reader, so that ample light could be gained 
without glare from the lamp itself, and the position of the light 
could be so adjusted as to avoid direct reflection from the paper. 
In small libraries with relatively small reading tables and a limited 
number of readers, this is probably the best arrangement. 

For reading rooms of large size having to accommodate many 
readers, movable lamps are somewhat troublesome, and if the table 
is not too wide fixed lamps along the center line, shaded somewhat 
as described and placed at a height to keep the light out of the eyes, 
can be made to do almost equally well. They should be so adjusted 


that the reader when sitting at the table will naturally place his 
book far enough under the lamp to avoid specular reflection into 
the eyes. Lamps adjustable in height, if not in position, are highly 
desirable for such reading-room use. Nothing better than the 
green-flashed porcelain shade has yet been devised for a table lamp. 

The distribution of light may sometimes be improved by placing 
a reflector within the shade directly over the lamp to widen the 
distribution, or sometimes by closing in the lower part of the shade 
with a ground-glass diffusing shield. The student lamp equipped 
with such a diffuser furnishes perhaps the very best form of illu- 
mination, where the use of movable lamps is permissible. 

Where many bookcases are around the reading room, lamps in 
opaque reflectors, with their apertures facing the shelves, may be 
advantageously used for lighting the books. An excellent form of 
trough illuminator for doing this work is shown in Fig. 125. The 
illumination here may well be 2 or 3 foot-candles. Similar lamps 
with reflectors are desirable for aiding the inspection of card cata- 
logues. In this case the illumination should be pushed somewhat 
higher, perhaps quite as high as at the tables, since the entries are 
not always clearly legible. 

In some instances very excellent results have been reached in 
reading rooms by general illumination only, sometimes in 'the form 
of wholly indirect lighting. Considering the area to be lighted and 
the usual height of such rooms, however, this method, while it may 
be made beautifully effective, is always wasteful of energy. 

A library stack can be very well lighted by the use of reflectors 
as just shown, and this is much superior to the common arrange- 
ment of bare lamps upon cords hung in the spaces between the 
shelves and turned on by anyone searching for books. The lights 
in each bay of a stack should be controlled by a switch at the 

Delivery rooms and similar public rooms are best treated by 
general illumination to the extent of 2 or 3 foot-candles, the exact 
form depending on the use to which the room is to be put. In 
case such public rooms contain any special features to be illumi- 
nated, the plan of lighting must be subordinated to these particular 
things. Here, even more than elsewhere, the installation of light 
sources must be considered as only a means to an end. Perhaps 
the very worst installation of lights in a library yet recorded was 
the original one in the delivery room of the Boston Public Library, 


a room enriched by the beautiful mural paintings of Abbey. Thia 
unique example of inappropriate lighting is shown in Fig. 144. The 
chandeliers were offensively obtrusive at best, and they were so 
located as to conceal the mural paintings by their glare instead of 
illuminating them so that they could be seen. The picture here 
shown was, of course, taken by daylight. By night the effect was 
almost indescribably bad. Fortunately this installation was soon 
thrown out and replaced by the effective and unobtrusive trough 
lighting system of Fig. 145. so inconspicuous as to be scarcely 

Fig. 144. 

noticeable in the cut, yet fully effective in revealing the beauty of 
the pictures. 

Perhaps the most troublesome portions of a library properly 
to illuminate are the newspaper and periodical rooms, where the 
reading matter is kept on slanting racks. With the lights as com- 
monly placed above these, there is almost a certainty of direct 
reflection from the paper into the eyes. Trough reflectors with 
diffusing screens may be used with advantage, and here, if any- 
where, the indirect system of lighting by the diffusion of the ceiling 
and walls finds its best application. It requires a considerable 
amount of energy to carry out the indirect lighting scheme in such 


a room, probably in the neighborhood of 1 candle power per square 
foot, but when properly done it averts completely the reflected glare 
from the paper while giving ample reading illumination. 

Museums and similar structures, like libraries, require somewhat 
specialized lighting, the main point being to illuminate the cases 
containing the objects on view to a fairly high intensity without 
producing disagreeable reflections from the glass surfaces and with- 
out placing lights where they will shine in the eyes of the visitors. 
Here again there is much to be said for indirect lighting, from sui- 

ng. 146. 

faces high enough to be out of the direct line of vision, and spacious 
enough to give very low intrinsic brilliancies. Certain cases will 
probably be found to require extremely brilliant lighting — more 
brilliant than they can economically receive from indirect sources. 
For such instances the familiar devices used in showcases are well 
adapted. These combined with a general indirect illumination 
probably furnish the best solution of the museum problems. 

In very large interiors without high galleries, arc lighting may 
be very effectively used, provided the arcs are well shaded. It is 
wise to group them so that no single arc shall entirely dominate 
the illumination at any particular point. It is better to lose a 


little in uniformity of the total illumination throughout the area 
than to take the chances of flickering, which is not entirely sup- 
pressed even in the best arc lamps. 

In a big space arcs can be treated much like incandescents in a 
small space, but the detail of the work varies so much that only 
very general suggestions can be given. Often temporary illumi- 
nation has to be undertaken, and must be fitted to the case in 
hand. One of the most beautiful examples of such work that 
ever fell under the author's notice was the illumination of Madison 
Square Garden for a chrysanthemum and orchid show some years 
since. The feature of this work was the very extensive use of both 
arc and incandescent lamps inclosed in Chinese lanterns. The 
huge lanterns containing the arcs were very striking, and the whole 
effect was most harmonious, while the illumination was thoroughly 
good. It is mentioned here merely as a clever bit of temporary 
lighting treated to suit the particular occasion. 

In this lighting of large interiors the smaller arcs worked on 
constant-potential circuits are very useful, although not very effi- 
cient. Those taking 5 to 6 amperes give excellent service, and 
fair results can be obtained with lamps working down even to 
4 amperes. Such arcs are equivalent to from 10 to 15 16-c.p. 
lamps in practical effect, and give a greater candle power per 
watt. The " intensified " arcs are by all means the best for 
such work. 

Incandescent lamps of the Nernst type or the largest sizes of 
tungsten lamps may be utilized in a similar way, in forming a good 
basis of illumination where the total amount of light is consider- 
able. In other words, when one is dealing with very large inclosed 
spaces, the lighting is simplified and made more efficient by utiliz- 
ing the mpre powerful radiants. 

|In certain cases, particularly railway stations and other build- 
ings likely to be rather smoky, arcs have to be the main reliance, 
since the globes of incandescents grow dim so quickly that cleaning 
them is an almost interminable job. Hence it is best to use 
comparatively few powerful radiants. The arcs should be carried 
rather high, at least 20 to 25 feet above the ground or floor. 
Assuming 0.5 foot-candle as the minimum, and taking into ac- 
count the illumination due to adjacent lamps, each arc can be 
counted on to illuminate over a distance at which it gives 0.25 
foot-candle. For close detail reference must be made to the actual 


illumination curves of the type of lamp used, and the general 
problem is analogous to street lighting. 

All arcs in inclosed places should have at least one opal globe, 
and when used where, as in railway stations, diffuse reflection is 
of small amount, should be provided with reflectors to utilize the 
light that would otherwise be wasted. 

Certain classes of interiors require, on account of the uses to 
which they are put, especial adaptations of the radiants, either 
in kind, amount, or position. One of the commonest demands is 
for an illumination of unusual brilliancy and steadiness in situ- 
ations like draughting rooms, and shops where fine work is done, 
where the eyes are under steady, if not severe, strain. Ordinary 
good reading illumination, such as we have been considering, 
must be considerably strengthened to meet these requirements. 
Simple increase in the number or power of the radiants sometimes 
meets the conditions, if such increase can be had without thrust- 
ing too powerful lights into the field of vision. 

It may be necessary to furnish 1 candle power for each 2 square 
feet of area, or, in extreme cases, 1 candle power per square foot. 
One of the most useful schemes for supplying such large amounts 
of light is the use of the inverted arc in connection with a very 
light interior finish. 

The ordinary continuous-current arc, in virtue of the brilliant 
crater of the positive carbon, throws its light downward; but if 
the current be reversed so as to form the bright crater on the 
lower carbon, most of the light is thrown upward toward the 
ceiling, and is there diffused. If, as usual, these arcs are arranged 
with inverted conical reflectors of enameled steel or the like, all 
the direct rays are cut off and the entire illumination is by the 
diffused rays. The result is a very soft and uniform light, white 
in color, and of any required brilliancy. Fig. 146 shows in dia- 
gram the principle of this device. In case a white ceiling is not 
available, large white diffusing screens over the lamps, of enameled 
metal or even of tightly stretched white cloth or paper, answer the 
purpose. Indeed, this was the original form of the device as shown 
by Jaspar at the Paris Exposition of 1881. 

With reference to Fig. 146, it is sufficient to note that the 
conical reflector should be rather shallow, just deep enough to 
throw the light wholly on the ceiling and upper walls, but shallow 
enough for two neighboring lights, as shown, to distribute light 



over each other's fields, which improves the average steadiness 
of the illumination. The arcs need no diffusing globes, a clear 
globe being sufficient, and open arcs may be freely used, to the 
material improvement of the luminous efficiency, never very high 
in this form of lighting. 

The heights of the arcs should depend somewhat on circum- 
stances regarding the appearance and the purpose of the lights, 
but will generally be half to three-fourths the height of the room. 
The reflectors may be from 3 feet to 6 feet in diameter, and may 
have an angle at the apex of 120 degrees to 140 degrees. Only in 
case of having to throw the light on special screens rather than on 



Fig. 146. — Lighting by Inverted Arcs. 

the natural ceiling should the reflectors have less aperture than 
just indicated. They then become of the nature of projectors, and 
the angle at the apex may be 90 degrees or so. 

As to the efficiency of such illumination, one may roughly assume 
1 watt per spherical candle power for powerful open continuous- 
current arcs, and may reckon on a loss of about one-half in the 
process of diffuse reflection. The diffuse illumination may then 
be taken as being in candle power about 0.5 the number of watts 
expended, not including artificial resistance. Thus, a continuous- 
current arc, taking 9 to 10 amperes at about 50 volts, utilized in 
this manner, will illuminate 250 square feet to 300 square feet on 
the basis of 1 square foot per candle power, or 500 square feet to 
600 square feet at 2 square feet per candle power. 


It must be noted that if ordinary inclosed arcs are used in this 
way, materially less light is obtained, as is well known. Even with 
both outer and inner globes clear, one cannot count on much better 
than 2 watts per mean spherical candle power. Alternating lamps 
require, of course, still more energy, and with inclosed arcs in 
.general one would hardly find it advisable to allow, when using 
ceiling diffusion, more than half to three-fifths of the area per watt 
just indicated for open arcs. Inclosed arcs have no crater, which 
operates greatly against their effectiveness in this class of lighting. 

These figures are necessarily only approximate, but while inclosed 
arcs have some conspicuous virtues, high efficiency as respects mean 
spherical candle power is not one of them. In all this lighting by 
diffusion the diffusing surfaces must be kept clean, else there will 
be much loss of light. Under even the best conditions, one does 
not do very much better than 2 watts per candle power, and lack 
of care or bad engineering may easily transform this into 3 or 4 
watts per candle power, which is no better efficiency than incan- 
descents would give. In point of efficiency the larger sizes of 
tungsten lamps are better than inclosed arcs for such work. 

The chief advantage of this diffused lighting is that it enables 
one to secure very brilliant illumination with white light, without 
trying the eyes with intense radiants. 

Such illumination has, however, one curious failing, in that as 
ordinarily installed it is shadowless, and the light has no deter- 
minate direction. For certain kinds of work this is a very trying 
peculiarity, severely felt by the eyes. It may be remedied in 
various ways, of which perhaps the simplest is the lateral dis- 
placement of the lamps shown in Fig. 147. 

This gives a predominant direction to the light, something akin 
to the effect produced by a row of windows along the side of the 
room, and is probably as near an approach to artificial daylight 
as can be attained by simple means. 

In using the arrangement of Fig. 147, about the same relative 
number of arcs is required as in Fig. 146, but they are placed in one 
row instead of two. The unilateral effect could be greatly enhanced 
by a diffusive screen db, Fig. 147, running, along back of the arcs. 
Its angle with the ceiling evidently should depend on the shape of 
the room. 

Unilateral illumination, whether diffused or not, is often desirable 
from a hygienic standpoint. In many cases well-shaded arcs may 



replace the diffused lighting just described, though such direct 
lighting is generally rather less steady. But it must be remem- 
bered that an arc having both inner and outer globes opalescent 
is scarcely, if at all, more efficient than incandescent lamps, assum- 
ing both to be worked off constant-potential mains; hence, unless 
the whiteness of the arc light is essential, incandescents, being 
steadier, are generally preferable. 

• In factories where colored fabrics are woven, and in shops where 
they are sold, white illumination is a matter of great importance, 
and arcs are especially useful. In the mills the necessary illumi- 
nation depends largely on the color of the fabrics. It should, as 

Fig. 147. — Unilateral Illumination. 

a matter of experience, range from 2 square feet per candle power 
to 1 square foot per candle power in passing from white to dark 
and fine goods. The candle power noted here is the mean spheri- 
cal, or hemispherical, if reflectors are used, taken from the real 
performance of the arc well shaded. This qualification means 
practically 300 square feet to 400 square feet for each arc of 450 
watts to 500 watts in the extreme case, and 600 square feet to 800 
square feet for white and light-colored goods. Shops where such 
goods must be sold by artificial light should be lighted on very 
nearly the same basis. For brilliant illumination, where color 
distinctions must be accurately preserved, the arc at the present 
time stands preeminent, and should generally be used. It must 


be remembered, however, that inclosed arcs are distinctly bluish 
unless the current is pushed up nearly to the limit of endurance of 
the inner globes, and hence, when used in situations where color 
is important, should have shades tinted to correct this trouble. 
The common opalescent inner globe is entirely insufficient for the 
purpose. The "intensive" arcs are best suited to the purpose. 

Where arc lights are not available, and it is desired to furnish 
approximately white light, there is difficulty in meeting the require- 
ment. Mantle gas burners, with extreme care in selecting tinted 
shades to correct the color of the light, may be made to give fair 
results, but are considerably inferior to arc lights. 

It should not be forgotten that good illumination in a workshop 
tends materially to increase the quantity and improve the quality 
of the work turned out. 

In most instances the color of the light within the range of 
ordinary illuminants is not a matter of considerable importance, 
but the light must always be reasonably steady. Hence the 
incandescent lamp and the mantle burner for gas are by far the 
most valuable sources of light commonly applicable. Ordinary 
batwing gas burners are probably the worst in point of steadiness, 
although a badly adjusted electric arc is a close second. 

Where very powerful radiants are desired, the large regenerative 
gas burners give a very brilliant and steady light. They throw 
out, however, a great deal of heat, which is sometimes objection- 
able, and are less economical of gas than the mantle burner. 
The modern "press-gas" lamps with inverted mantles are still 

A very special branch of illumination is the lighting of immense 
inclosed spaces, such as are found in exposition buildings. This 
work is on such a large scale that it almost partakes of the nature 
of outdoor lighting, with which it is very intimately connected 
as a practical problem. The amount of light required in single 
inclosed spaces of colossal dimensions, like exposition halls, varies 
considerably according to the practical use to which the space is to 
be put. As a rule, the most brilliant and useful illumination in 
these large spaces is secured by the use of arc lights to the exclu- 
sion of other illuminants. In a building covering one or several 
acres, and perhaps 100 feet or more in height, incandescents of 
ordinary powers look lost; and if the roof is not to fade away into 
darkness, a very large number of lights must be required to bring 


it into prominence, placed so high from the floor as to be of little 
service for the general illumination. 

Moreover, such buildings have generally a large amount of 
glazed side and roof space, furnishing the ordinary daylight illu- 
mination. Consequently the walls and ceiling diffuse very little 
light. With arc lights the power of the individual radiants bears 
some respectable proportion to the size of the space to be illumi- 
nated. The luminous efficiency is increased, and, by sufficient 
massing of lights with reflectors, even the highest halls can be 
admirably lighted. The work can, of course, be beautifully done 
with incandescents if enough are available, but at considerably 
lessened economy. 

The amount of light required per square foot of floor space is 
very considerable, owing to the height and bad diffusing proper- 
ties of the building, and for the best results 1 actual candle power 
should be furnished for each 2 square feet to 3 square feet, accord- 
ing to conditions. 

Incandescent lamps have a very high decorative value in con- 
nection with such work, but to be used effectively must be massed 
somewhere near the plane of illumination, lights in and about the 
roof being practically only for decorative purposes. Used in suffi- 
cient numbers, however, they give, in virtue of their complete 
subdivision of the illumination, a better artistic result than can 
be obtained with arcs. 

The subject of exposition illumination is so large and so special 
in its character as to be hardly appropriate to the scope of the 
present work. 



J3y exterior illumination is here meant that which is applied out- 
side the confining walls of buildings. Interior illumination, which 
is circumscribed by such walls, is powerfully modified by their par- 
ticular characteristics as to color, texture, and coefficient of reflec- 
tion. Of the six surfaces which bound a typical interior space, 
four or five are generally moderately good reflectors, or at least are 
not so low in reflectivity as to be at all negligible. One, commonly 
the floor, is often nearly or quite negligible, and sometimes, in the 
case of high vaulted ceilings, another bounding surface may be left 
for the most part out of account. In exterior lighting the case is 
radically different . In some instances there are no bounding sur- 
faces to the space illuminated of such character or at such distances 
as to afford any secondary illumination worth mentioning. In 
other cases there may be two or three reflecting surfaces, generally 
rather bad, to be considered, but in all cases the upper limiting 
surface is absent aftd the condition generally approximates the 
illumination of an indefinitely extended room with a poorly reflect- 
ing floor and an absolutely black ceiling. 

One, therefore, deals, in exterior lighting, chiefly with light re- 
ceived directly from the radiants, and in so far the case is theoreti- 
cally simpler than interior lighting. On the other hand, the lower 
bounding surface in exterior lighting may be relatively important, 
particularly in certain cases of low illumination to be described 
later on. Now and then there are lateral bounding surfaces which 
are not negligible, and there are also extraneous sources of light 
which in practical illumination are of great importance, but, on the 
whole, exterior illumination depends for its effective magnitude 
upon light received directly from the radiants in use. 

From the economic standpoint exterior illumination presents a 
favorable case, inasmuch as relatively low intensities are employed, 
since space out-of-doors does not need to be lighted to the degree 
required for occupations or amusements customarily carried on 
indoors. Broadly, then, the art of exterior illumination deals gen- 



erally with the distribution, directly from one or more radiants, of 
a moderate degree of illumination without much effective aid from 
any secondary sources of light. 

Most generally, illumination out-of-doors is applied to a single 
surface, the ground; but there are cases in which the fundamental 
requirement is the lighting of vertical surfaces, such as are pre- 
sented by buildings. It is this class of lighting which, perhaps, 
bears the greatest resemblance to the conditions of interior lighting 
and which will be considered in the next chapter. 

The main class, therefore, of exterior illumination here to be 
considered, and the one of the greatest economic importance, is 
that of street lighting, in which the distribution has to be chiefly 
lengthwise of the streets. Bounding surfaces in the form of build- 
ings may or may not be of material importance, and the intensity 
required is rather moderate. It is lighting in one dimension, rather 
than in two, as in the case of public places and parks, or in three, 
as in the case of interiors. Prior to discussing this, however, it 
will be well to consider the somewhat more general problem of place 
and park lighting. 

The lighting of public places and parks differs from street light- 
ing in that the areas to be illuminated are not narrow strips like a 
street, but extend in both directions, and in the case of public 
squares the lighting of the adjacent buildings is a thing not to be 
left out of account. The purposes of these kinds of lighting differ 
very widely. Public squares are illuminated with special reference 
to the convenience and pleasure of the people who use them, often 
in great numbers. Such places are frequently dense centers of 
traffic along the streets that meet upon them, are generally located 
in the more thickly populated parts of the city, and are often scenes 
of great activity during at least the earlier hours of the evening. 
• Man has become steadily more and more a nocturnal animal, and 
it is in these public squares that provision must be made for his 
habits. Both his protection and his convenience are objects which 
must be borne in mind when designing the illumination. The police 
value of lighting has long been recognized, and emphasis was laid 
on it in an interview recently by the Chief of Police in Paris, who 
pleaded for adequate all-night lighting as an adjunct for the preser- 
vation of order. In considering public squares, the value of ample 
light as preventive of crime is very considerable, but perhaps less 
important than it is in some of the streets. A public square is 


not a spot generally chosen for "holdups" or other extreme crimes 
of violence. It is, however, a location where the pickpocket and 
petty thief may ply their vocations, and for full protection against 
these gentry good illumination is needed. 

Fundamentally, the lighting of a public square is for the con- 
venience of the passersby. They not only wish to walk without 
tripping over obstacles, or drive without plunging into open man- 
holes, but they wish to meet and recognize their friends without 
bumping into them, to glance at a railway time-table, to read the 
address on a letter or the number on' a house, and, in general, to 
see as comfortably and get about with as little thought of incon- 
venience from lack of light as would be the case toward the end 
of a winter afternoon. In other words, the peculiar requirements 
of convenience demand that public squares which are largely used 
should be liberally lighted — as well lighted as the best-lighted streets, 
much better lighted than the ordinary streets. It is consequently 
necessary that they should be lighted with some approach to uni- 
formity, otherwise there will be dark spots not only unpleasant in 
effect but inconvenient for the man on the street. From a practical 
standpoint such requirements can be met thoroughly well in only 
one way, — by the use of a very large flux of light from sources 
placed high enough to be out of the immediate field of view. This 
is akin to the ordinary requirements of interior lighting, in that 
one should be able to see easily and comfortably without brilliant 
sources of light intruding themselves in the direct line of vision. 

The lights ordinarily used for street lighting, if sufficiently 
numerous to give the requisite volume of illumination in a public 
square, are certain to interfere with vision by their brilliancy and 
position near the line of sight. The author calls to mind three 
famous places which serve as examples of the bad and good meth- 
ods of place lighting. One of these is the Place de la Concorde, 
Paris, lighted with innumerable small units placed on short posts 
that stand in serried ranks all about the famous spot. The light- 
ing of the pavement is moderately bright, but the effect is dis- 
tinctly unpleasant and inadequate; petty from the great number 
of lamps and the obtrusiveness of their supports. The second is 
Trafalgar Square, London, lighted with arcs to a somewhat higher 
degree than the Place de la Concorde, but yet missing something 
of distinguished beauty or notable excellence in the results. It 
is a fairly well-lighted square, which could be made much better 


were the lamps placed farther out of the field of view and the 
total volume of illumination considerably increased. Finally, as 
an example of the very best that has been done in such lighting, 
one may mention the western approach to the Brandenburger 
Thor in Berlin, which is brilliantly and beautifully lighted by two 
groups of enormously powerful lamps placed more than 20 meters 
high on columns which are works of art by day as well as by 
night. Fig. 148 shows the daylight aspect of the place. The 
actual illumination on the pavement, while amply brilliant, is 

Fig. 148. 

probably no higher than is reached in Copley Square, Boston, or 
in any one of several public places in other American cities. 

The design of the illumination in a public square is not a simple 
matter. First, considering the amount of light required in order 
to meet the requirements of being able to read notes, time-tables, 
and addresses comfortably, as well as to recognize persons quickly 
and easily, the illumination must be pushed far beyond that found 
in most American streets unless almost under the lamps. To 
meet these requirements the average value of the effective illu- 
mination should be not less than 1 meter-candle and the minimum 
should be at least 0.5 meter-candle. Anything less than this is 
insufficient for the purposes mentioned, and more is preferable. 


One can form a cursory idea what this intensity means by 
realizing that full moonlight is in our latitude on a clear night 
about 0.3 meter-candle, a degree of illumination that reduces 
visual acuity to about 0.25 or 0.30, as shown by actual experi- 
ment in moonlight, and- reduces shade perception in a similar 
degree. Both the loss of acuity and the increase of Fechner's frac- 
tion below 1 meter-candle are very rapid, and at these low illumi- 
nations the eye is peculiarly susceptible to the effect of bright 
lights within the field. 

The term " effective illumination" is used advisedly, with full 
knowledge of the fact that there is some discussion as to what 
constitutes effective illumination for the purpose of lighting such 


Fig. 149. 

a space as we are considering. While the illumination on a sur- 
face normal to a ray from a radiant of known power follows simply 
the inverse-square law, if the ray does not fall normally upon the 
surface the intensity is reduced in proportion to the cosine of the 
angle of incidence for a horizontal surface, and in proportion to 
the sine of the same angle for a vertical surface. Consequently, 
if one attempts to reckon the illumination to be received at a 
particular point in a public square, he finds himself in a quan- 
dary as to whether he shall reckon the illumination as on a 
normal plane, the illumination resolved on a horizontal plane, or 
resolved on a vertical plane, the three hypotheses leading to three 
radically different results as to the value of the illumination. 
The curves of Fig. 149 give the three values of the illumination 


obtained on these three hypotheses from a source of 1000 uniform 
spherical candle power placed at a height of 10 meters. Which 
of these divergent values should be reckoned as the correct one 
for the purpose of designing illumination? The question is an 
intricate one on account of the varying purposes for which one 
requires light in such a situation. 

Here, again, the similarity to interior lighting becomes evident, 
since the case corresponds quite closely to that of a room lighted 
from several sources. The solution is, in the author's judgment, 
indeterminate, since there are more unknown and perhaps un- 
knowable quantities than definite data which can be applied to 
them. One can, however, arrive at a common-sense approximate 
solution by establishing this criterion, — that the light shall be such 
as to meet the severest practical test among the various require- 
ments of its use; that is, the reading test. For this one can always 
readily take advantage of normal illumination, and one customarily 
does so. This requirement means, therefore, that the normal illu- 
mination received from the nearest light shall at no point fall below 
0.5 meter-candle, and shall, as a whole, equal or exceed 1 meter- 
candle. With this quantity of illumination all practical require- 
ments other than reading are met very easily. 

The problem of design, then, resolves itself into a comparatively 
simple construction, — the placing of radiant sources so that if one 
draws about each of them a circle at such distance that the nor- 
mal illumination received from the source at that circle shall be 0.5 
meter-candle, these circles shall overlap so as to fully cover the area 
concerned. The subsequent design consists in so planning the dis- 
tribution from each source that its effective radius of action shall 
be as great as possible. With all practical illuminants the illumi- 
nation, if sufficient at the periphery of the circle, will be sufficient 
for all points within. 

For the purpose in hand, the fundamental equation connecting 
the various quantities is 

L n =^cos 2 a, (1) 

where L» is the illumination, I the height of the radiant above the 
plane of reference, J the intensity, and a the angle of incidence, 
which is equal to the angle between the ray and the lamp-post. 
Of these quantities in actual computation any one may be assumed 
on the conditions, or any one may be required to be found. J and 


the angle a are dependent variables, and in practice are taken 
from the distribution curve of the radiant. This being known, the 
required height of the lamp to produce a given illumination, L n , 
can be obtained from the transformed equation, 

p = ZC08»a (2) 

L n 

For instance, taking L» at 0.5 meter-candle and J for the angle of 
incidence 70 degrees, as 2000 candle power, I comes out at about 
22 meters; and it will be generally found that with distributions 
common for powerful illuminants the heights, for illumination of 
the order of magnitude here required, come out rather large, higher 
than it is generally convenient to place the lamps. 

Again, the height of the lamp being chosen at some easily prac- 
ticable figure and the curve being known, the angle of incidence 
corresponding to the required illumination is given by the trans- 
formed equation, 

cos 2 a = -j- • (3) 

Whether the angle of incidence is assumed, or thus reckoned, the 
radius of the circle for the required illumination at the periphery is 

r = I tan a. (4) 

Since a and / arc mutually interdependent, the solutions thus ob- 
tained are not exact; but having the distribution curve of the lamp, 
a slide rule, and a table of natural trigonometrical functions, one can 
get at the facts in the case in very short order. As an example 
in the application of these formulae, the following data derived from 
the illumination of Copley Square, Boston, which is lighted by four 
very powerful flaming arcs, may be instructive. Here I equals 16 
meters and / is very conservatively taken at 2000 for angles in the 
vicinity of those dealt with. Fig. 150 shows the curve of the lamp 
with opal globe. Applying equations 3 and 4 for L» equals 1 meter- 
candle and 0.5 meter-candle, respectively; r equals 41.6 meters for 
1 meter-candle and 62.5 meters for 0.5 meter-candle, approximately. 
Fig. 151 shows these circles as laid down on a map of the Square. 
It will be seen that the 1-meter-candle circles overlap liberally, 
and the 0.5-meter-candle circles almost touch the adjacent lamps. 
It was considered desirable here, especially on account of the fine 
neighboring buildings and the large traffic through the streets, to 



carry the illumination high, and the 0.5-meter-candle circles reach 
well out into the adjacent streets. The great overlap of the circles 
of illumination renders the lighting extremely uniform, and one can 

Fig. 150. 

read a newspaper anywhere in the Square without any sensation of 
glaring brilliancy being perceptible, owing to the great height of the 
lamps. Ordinarily the 0.5-meter-candle circles in place lighting 

♦ FlMQAAtt 

Fig. 151. 

would overlap about as much as the 1-meter-candle circles do in 
this instance. Had it been feasible to use poles fully 20 meters in 
height, a slightly different reflector could have been advantageously 



i on the lamps with the probable result of increasing the 
efficiency of the lighting very materially. But the lamps being 
on series circuits, on which the use of iron poles is not permitted 
in Massachusetts, it was not practicable to go higher. Fig. 152 
shows the night view of Copley Square thus illuminated. Applica- 
tions of the principle of design here suggested are independent of 
the power or character of the radiants and will serve for the light- 
ing of public places of any size or importance to any degree of 

Fig. 153. 

brilliancy. In places of modest area a single central fixture bearing 
a group of powerful lights, or even a single lamp of high power, 
may yield admirable results. Fig. 153 shows an excellent example 
of such practice in the Alt-markt at Dresden, where the central 
feature is the great ornamental post bearing its cluster of six flam- 
ing arcs. 

The lighting of parks differs somewhat radically from that of 
other public places for the simple reason that most parks are so 
little used after nightfall, except in very limited portions, that any 


considerable degree of illumination is unnecessary. Now and then 
one finds a park which is used freely in the evening, and in such 
cases lighting on a liberal scale ought to be supplied, rising, rarely, 
to that appropriate for other public places. Generally speaking, 
however, the purpose of park lighting is purely the preservation of 
order and the marking of what are, so to speak, thoroughfares 
through the park. 

From the police standpoint, which is the important one in park 
lighting, the requirement is for moderate illumination without dark 
spots in which the disorderly can lurk. Hence, as a rule, powerful 
radiants which, unless brilliant illumination is attempted, would be 
widely spaced and would tend to cause somewhat dense shadows 
ought always to be avoided in park lighting. Their only proper 
application to such work is where the illumination should approxi- 
mate that of other public places, and in the case of large open 
spaces. Parks in general, therefore, require less light than any other 
class of public spaces which require illumination at all. In many 
instances, where parks are large and wooded, there can be no 
attempt at a general illumination even for police purposes, except 
in certain spots and along certain routes through the park. Where 
lighting is attempted at all, its intensity along the ways in the park 
should be the same as in a very moderately lighted street. On 
such ways it may fall still lower provided it is reasonably uni- 
form,— as low, indeed, as average moonlight, perhaps 0.2 meter- 
candle or thereabouts. The objects to be seen by such lighting 
being nothing smaller than persons, the demand for visual acuity is 
small. Practically the problem amounts to furnishing enough light 
in a certain area to prevent unwarranted persons from lurking in 
the park after nightfall. Any light, therefore, by which the wander- 
ing policeman can make out a figure serves the purpose. 

As a result of this police requirement the distribution of lights 
in a park has sometimes to be very singular, the lights being placed 
utterly irrespective of any systematic order, but where they will 
abolish dark spots under trees and behind shrubbery. For this 
purpose the lights are preferably of only moderate power, and 
should be placed low, where they can shine below the branches 
of trees. It is also important that lights so placed should be 
thoroughly screened so as to avoid glare. Under the conditions 
required the guardians of the peace can fulfill their functions most 
successfully when their eyes are adapted to a dim light, and dark 


adaptation is spoiled by a very brief exposure to a powerful light 
or one of high intrinsic brilliancy. Even incandescent lamps, 
whether gas or electric, used for such park lighting as is here under 
consideration, should be put in ground glass or opal shades, pref- 
erably the latter, so that their light-giving power may be utilized 
without interfering with the vision of those looking toward them 
or passing near them. 

The best results in park lighting in the writer's experience have 
been with 100-candle-power tungsten lamps in 12-inch diffusing 
balls, mounted about 10 feet above the ground, and in positions 
designated after conference with the police authorities. Larger 
units than these can rarely be utilized to advantage on account of 
their being too bright, and small ones similarly installed can fre- 
quently be made to serve the purpose. If similar lights are installed 
along the ways in the park which are desired to be lighted, they 
will do excellent service when so spaced as to give illumination 
similar to average moonlight, say 0.2 to 0.25 meter-candle. Where 
people congregate in the park the illumination should be carried 
higher, up to at least 0.5 meter-candle. In open spaces arcs can 
here be made to do good service, the illumination being planned 
exactly along the same lines as in the case of the public places 
already treated. Small units closely placed are less effective, ex- 
cept in lighting spaces like open-air restaurants, in which the lights 
should always be shielded by diffusing balls or shades. The mini- 
mum intensity of light in such places should be at least 0.5 meter- 
candle, enough to enable one to read a menu card or program. 
Park lighting, therefore, would seem to belong in a special class 
as regards intensity and distribution, and from its low intensity 
requires that particular pains should be taken to avoid glare. 

Street lighting is in its origin and development essentially a police 
measure. Its history goes back to mediaeval times, in which the 
streets, mostly unpaved and wholly undrained, were bad enough 
by day but worse by night. They were infested by thieves and 
highwaymen, cut-throats and drunken roisterers with rapiers ready 
for a quarrel. Paris, in particular, in which we have the first 
records of street lighting, was the scene of constant brigandage 
and crime from almost the earliest days of which we have record. 
Early in the fifteenth century, under Louis XI, flambeaux were 
ordered at the street corners, and lanterns in the householders' 
windows to cooperate with the night watch in promoting public 


safety. Over and over again for the next two centuries and a half 
such ordinances were reiterated, always openly on the score of 
public safety from crimes of violence. In 1558 lanterns were 
ordered, at the corners of streets and at other suitable places, to 
be kept burning from 10 o'clock in the evening till 4 o'clock in the 
morning through the winter months. By these dim and flickering 
lights the course of the streets was at least marked, but they 
afforded scanty protection against marauders, and those who could 
afford it went accompanied by a retinue of torch bearers and an 
armed guard when traversing the streets at night. It was not 
until more than a century later that anything approaching public 
lighting was seriously attempted. In 1662 the first public lighting 
concern was given a franchise under a royal edict of October 14. 
This was a private enterprise of the Abb£ Laudati, and its chief 
feature was the constitution of a corps of public lantern bearers 
carrying lanterns or flambeaux of a specified size and bearing as 
insignia the arms of the city. These were stationed at fixed posts 
along the streets and for a small fee would escort the nocturnal 
wanderer more or less safely on his way. Systematic street light- 
ing was inaugurated about five years later, and the effect on 
public order seems to have been immediate, for at least two medals 
were struck within the decade, celebrating the institution of public 

London was still worse off and save for the transitory effect of 
ordinances requiring householders to hang out lanterns the streets 
were unlighted and almost as full of danger as those across the 
Channel. It is worth noting that about this period some ignoble 
soul, whose name has very properly perished in oblivion, devised 
the original moonlight schedule as a measure of poor and pitiful 
saving. It was tried first, probably, in Paris, where it was railed 
against as "candle-end economy." London copied from Paris, 
and had made little progress before the end of the seventeenth 
century. At this period, however, the chief streets of Paris were 
systematically lighted by lanterns swung across the street, still the 
most efficient position for street lights. They were placed at about 
20 paces apart and hung some 20 feet above the ground or pave- 
ment, as the case might be. Fig. 154 shows a nearly contempo- 
raneous view of such suspensions. It was well into the eighteenth 
century before street lighting at public expense was customary even 
in the capitals of Europe. The subsequent 200 years have seen an 


immense change in methods and material, but the purpose of street 
lighting has remained the same, and it is now, as it was four 
centuries since, a measure of public safety and an adjunct to the 
police force. 

For practical purposes the street lighting of the seventeenth and 
eighteenth centuries was altogether insufficient and ineffective, but 
despite this it was found, as it is found to-day, a great preventive 
of crime. As the use of the streets by night has increased the 
necessity for lighting has grown with it. Lighting to-day bears 
a closer relation to public safety than it did when the only occu- 
pants of the streets after nightfall were a few crawling carriages 

and a few belated pedestrians. It is necessary not only to light 
the streets well enough to mark their course and serve for the 
assistance of the guardians of the peace, but well enough to dis- 
tinguish the way clearly, to avoid obstacles even when going at 
fairly high speed, to distinguish and recognize persons, and to tell 
where they are and what they are doing. The police should be 
able to note the actions of suspicious characters before they stumble 
over them, or to detect the number of a law-breaking automobile 
before it has vanished into the distance. All these requirements 
of a complicated civilization demand lighting upon a vastly more 
liberal scale than sufficed for earlier days, or than is found in 
many localities even now. It is pertinent, then, to inquire into 
the conditions of visibility that are present with artificial light in 


the streets, and to find their bearing upon the intensity of light 

Except in the vision of details at comparatively short range, we 
see things in virtue of their differences of color and of luminosity. 
In weak light color as such is inconspicuous, so that practical vision 
depends chiefly upon the power of distinguishing differences of 
luminosity. So far as the problems of artificial illumination are 
concerned, objects do not range over a wide scale of luminosity. 
Whatever may be the absolute values of light received, the relative 
values as expressed by the coefficient of reflection range practically 
from about 0.8 to 0.01, or a little less. In other words, the blackest 
object returns about one-eightieth the light returned by the 
brightest object. Ability to distinguish between stationary objects 
by their difference in luminosity depends, then, on the capacity of 
the eye as regards shade perception. The fundamental fact under- 
lying this is that the eye can perceive within a wide range of 
absolute intensity a fairly constant fractional difference in lumi- 
nosity. In bright light it ranges, say, from 2 to 0.5 per cent, with 
modest variations under special circumstances both ways from 
these values, which hold measurably well for values of the illumi- 
nation from about 10 meter-candles up. As the illumination falls 
below this point there is a material increase in Fechner's fraction 
under ordinary circumstances, and we see less well, so that by the 
time the illumination is down to 1 meter-candle our shade percep- 
tion is very seriously impaired, as is also our ability to distinguish 
details, — visual acuity. It should be mentioned that the loss 
of shade perception at low illuminations is very powerfully influ- 
enced by the state of adaptation of the eye with respect to light 
or darkness. With the eye well adapted to the dark, fairly good 
shade perception can be carried to illuminations very much lower 
than ordinary. In fact at 1 meter-candle or a few tenths the 
value of Fechner's fraction is influenced very much more by the 
state of the eye as regards dark adaptation than by anything 
else, so that when one is seeing fairly under a very low illumina- 
tion, anything which tends to spoil the dark adaptation produces 
immediate blinding with respect to things otherwise easily seen. It 
is chiefly this fact which, from the standpoint of street illumina- 
tion, renders glaring lights so troublesome. 

Referring these things to the physiology of vision, the situa- 
tion may be summarized by saying that below an illumination of 


1 meter-candle, normal daylight vision, which is chiefly associated 
with the cones of the retina, is rapidly failing and throwing the 
burden of vision upon the rods. 

There is, then, a physiological dividing line that can be drawn 
between illumination which permits fairly good seeing and illumina- 
tion which leaves only the residual twilight vision; between the 
illumination which enables one to perceive things with some degree 
of definiteness and that with which one perceives chiefly forms and 
shadows. The exact position of this line is somewhat difficult to 
define, as it varies more or less in different eyes and under different 
conditions. At 0.5 meter-candle one has certainly not passed fully 
into conditions of twilight vision. Color perception, though much 
impaired, has not disappeared, and acuity, though failing, still 
remains in sufficient degree to permit casual reading, although with 
some little difficulty.* At 0.1 meter-candle a condition is reached 
where one depends almost entirely on rod vision. Acuity has been 
enormously reduced, and shade perception has become almost 
wholly dependent on dark adaptation. The point at which cone 
vision goes rapidly out of service, and rod vision as rapidly takes 
its place for what it is worth, is somewhere about 0.2 or 0.25 
meter-candle, and daylight vision is not very dependable below 
0.5 meter-candle. We have here, then, the physiological charac- 
ters of the eye which are already well determined by investigation 
directed particularly upon them, as the basis of a physiological 
criterion of illumination. In twilight vision one sees things not as 
distinctly perceptible, but as dim forms and shades of uncertain 
boundaries and character. Only when the objects subtend a fairly 
large visual angle does one see them in the least clearly. This is 
the familiar vision of a bright starlit night or a dimly illuminated 
street. In its beginnings one cannot even distinguish large station- 
ary objects from their background. The first perception is that 
of objects in motion, which seem to catch the eye more readily 
than when they are at rest. This is a familiar phenomenon in 
trying to pick up objects with a night glass at sea. They can be 
caught by sweeping when they quite escape detection on apparently 
slower and more careful search. This, too, is probably characteris- 
tic of the vision of nocturnal animals. 

The next stage of vision presents objects either as vaguely sil- 
houetted against a lighter background or as faintly lighted against 

* One can, for instance, still read by the light of a candle 1.4 meters distant. 


a darker background. By further increase of lighting some details 
begin to be perceptible, and when the illumination has passed the 
critical point of 1 meter-candle or a little more, to which reference 
has been made, further details come into view and objects take on 
a more natural aspect. The interesting theory of silhouetting as 
a feature of street illumination, which has been recently advanced, 
really concerns chiefly twilight vision and emphasizes the desir- 
ability under this condition of having a light background, not be- 
cause one can see a dark patch on a light background any better 
than a light patch on a dark background, but because many things, 
and particularly the large things, which alone fall within the scope 
of twilight vision, are themselves commonly rather dark in surface, 
and, consequently, are not easily rendered lighter than the back- 
ground. On the other hand, many objects are of surface lighter 
than, say, an asphalted street, and, consequently, are seen as 
light objects, while commoner than either condition is that of 
seeing an object in twilight vision only by its shadow, as one 
sees a distant pebble in the street in the beam of an automobile 

No illumination which depends chiefly on twilight vision can be 
seriously considered for the important purposes of street lighting. 
It has its useful place merely in enabling one to find the way. 
To be effective for the purposes of ordinary traffic, or as an adjunct 
to proper policing of a city, illumination must be sufficient to estab- 
lish, at least to moderate extent, the conditions of cone vision. 
The wayfarer wants to distinguish the shadow of a post from a hole 
in the pavement before he is fairly upon it. The man who is 
driving along a street needs to see his way clearly without risk of 
running into the gutter, and the policeman should be able to tell a 
belated householder from a burglar using his jimmy on a front 
door. And, finally, in many places it is highly important to have 
enough additional light to distinguish faces readily, to see even 
trivial obstacles easily, and to read the numbers on houses or, if 
need be, consult an address book or a time-table. These con- 
siderations lead inevitably to the conclusion that unless one is 
prepared to meet the most exacting conditions of street illumina- 
tion throughout the city he must be willing to classify the lighting 
that is to be undertaken, and to light each street according to its 
needs, bearing in mind the amount and kind of nocturnal traffic, 
and particularly the requirements of public order. We may here 


consider streets as divided for the purpose of lighting into first, 
second, and third classes. 

By first-class streets we mean the chief streets of a city from 
the standpoint of amount of nocturnal traffic and the requirements 
of the police. A chief street may be the principal business street 
of a city, that is humming with activity after nightfall. It may 
be a street leading to a crowded railway station where carriages 
and foot passengers are constantly circulating until late in the even- 
ing, or it may be a comparatively humble business street in a por- 
tion of the city where the police have found from experience that 
only constant watchfulness can keep down crime. 

The ordinary streets of a city or town fall into another category. 
Traffic after nightfall is light or only moderate. The streets are 
reasonably orderly and the general conditions are such that neither 
from the viewpoint of the wayfarer nor from that of the policeman 
is brilliant illumination necessary. Such streets are the ordinary 
quiet residence streets of the average city and the business streets 
on which there is little traffic by night. These streets usually 
would figure up to two-thirds or three-fourths of the total mile- 
age in the average city. These may be regarded as second-class 
streets from the illuminating standpoint, requiring good lighting, 
but not of the highest pitch of brilliancy. 

Finally, there are, in every city, a number of streets which re- 
quire practically very little illumination. They are mostly in the 
outlying portions of the city, sometimes scantily built up, and some- 
times they are mere roads leading away from the structural part 
of the city, but still within its jurisdiction. For such streets it is 
necessary to provide only such illumination as will serve to mark 
the way and to render progress through them easy considering the 
conditions of traffic. 

Occasional outlying streets, not at all important as residence 
streets, are still considerable traffic carriers, being through-roads 
from one part of the city to another, or from the city to some 
particular suburb or neighboring town. Such, from the standpoint 
of the illuminating engineer, are second-class streets rather than 
third-class streets. They demand the illumination required by 
considerable traffic. It is difficult to lay out any exact criteria for 
this classification, but there is no chief of police who could not, 
after a little reflection, make it with practical precision. 

As to the intensity of lighting required for streets of these several 


classes, the requirements have, by popular consent, been slowly 
and steadily rising. On the physiological basis which enables one 
at least to determine what lighting is necessary to reasonably 
good vision for various purposes, one can form a fair approxima- 
tion of conditions to be met. First-class streets in constant use for 
dense traffic of one kind or another, or so classified from the police 
standpoint, certainly require reading illumination, and this is the 
kind of illumination that first-class streets get in most European 
and some American cities. The intensity of the illumination re- 
quired is practically that already specified for public places, that is, 
an extreme minimum of at least 0.5 meter-candle, an average of 
fully double that amount. Theory and practice concur in holding 
that a street so lighted is well lighted. The same question arises 
here as regards the way illumination should be reckoned that was 
answered concerning the lighting of public places, and the answer 
is much the same for both cases. In each case it should be under- 
stood that the minimum cannot be permitted to apply to any con- 
siderable portion of the street area. The intensity here specified is 
substantially that deemed advisable by several foreign investigators 
and carried into practice with entirely satisfactory results. 

As regards second-class streets the requirements are, of course, 
less severe. There should be, as a matter of convenience, light 
enough to recognize a friend without stumbling over him, to read 
an address, or see the number of a house comfortably. The 
average illumination for this purpose may be set at not less than 
0.5 meter-candle, and the minimum should be high enough not to 
drive one into the physiologically undesirable condition of relying 
upon rod vision only. This would imply that the minimum should 
be nowhere less than about 0.25 meter-candle. Streets so lighted 
will be comfortably bright near the lamps, and the lighting will 
be as good as moonlight even at the darkest spot. This degree 
of illumination is excellently serviceable for the majority of second- 
class city streets. 

Finally, we come to the third-class streets. If there is light 
enough to mark well the way and to disclose persons or vehicles 
in ample time for one to avoid them, it is sufficient. The degree 
of illumination required need not be greater than is afforded by 
bright moonlight, and should be fully as great as one finds in rather 
dim moonlight. It is illumination similar to that which is required 
for some of the park lighting, to which reference has been made, 


ranging, say, from 0.30 down to 0.1 meter-candle, or thereabouts. 
At such low intensities it would be better to cut down the 
intrinsic brilliancy of the lights by screening, $s in park lighting, 
so as not to spoil the dark adaptation which is necessary for 
utilizing so low a degree of light. In streets so dimly lighted the 
silhouette effect is rather marked, and it is not desirable to forego 
the advantage of a tolerably light surface on the street. The 
coefficient of reflection of roadways varies greatly according to their 
surface and the angle of incidence. At fairly large angles of 
incidence a dirt road or a dusty bit of ordinary macadam may 
give' coefficients as high as 0.25 to 0.35; under similar conditions 
a dark pavement or a bit of oiled macadam may give coefficients 
from half of these figures down to as low as 0.05, which low values 
greatly increase the difficulty of proper illumination. 

Considerations of economy in street lighting enforce such classi- 
fication of streets as here described. Few cities can afford even 
at the present scale of public expenditure to light brilliantly any 
large proportion of their streets. If an attempt were made to light 
all the streets alike, there would be no first-class lighting at all. 
In small cities where the traffic is never very dense, and the use 
of the streets at night moderate, very little first-class lighting is 
required, and the amount necessary will diminish with the traffic. 
The burden of lighting, perhaps, falls more heavily on small cities 
than on large, owing to the large amount of street mileage compared 
with the assessable values. Hence, in such places, there will 
be, and properly may be, from the conditions, a relatively con- 
siderable amount of third-class lighting, but even so the cost of 
lighting is sometimes a serious matter. To keep down expense 
and yet to adjust the lighting conditions as well as may be, various 
attempts have been made to reduce the hours of lighting per year 
while yet meeting fairly the practical requirements. The earliest 
attempt of this kind, to which reference has already been made, 
was based on cutting out all the lights on moonlight nights. This 
scheme is apparent in the various moonlight schedules which have 
been used. Such schedules are all unsatisfactory, for the reason 
alleged against them from the beginning, — that the weather is no 
respecter of moonlight, and the nights near full moon are, in point 
of fact, sometimes as dark as the darkest. The only suitable 
lighting for cities of any importance is the all-night and every- 
night schedule. This is commonly based on starting the lamps 


half an hour after sunset and extinguishing them half an hour 
before sunrise every day in the year. This amounts to a total of 
nearly 4000 hours per year. For any given locality this should 
obviously be based on local time and not on standard time. The 
intervals between sunset and lighting, and extinguishment and 
sunrise, are subject to some modification in the practice of various 
cities, changing with the season of the year, but the all-night and 
every-night schedule will be found to run between 3900 and 4000 
hours, seldom being less than the former or exceeding the latter. 
If the full schedule as first suggested is to be modified at all, it 
is better to modify it in the morning hours than in the hour of 
lighting up, by reason of the greater traffic in the evening. 

The so-called moonlight schedules vary considerably according 
to the tacit assumptions made regarding the effectiveness of 
moonlight, but run commonly a little over 2000 hours per year. 
A modified moonlight schedule, as used in a number of cities, 
starts with lighting from dusk to midnight every night, and takes 
on the moonlight schedule by extinguishment approximately an 
hour after moonrise after midnight. Such schedules run to about 
3000 hours per year. The reduction in cost is not, of course, pro- 
portionate to the reduction in hours, so that the economy is to a 
certain extent rather apparent than real. A better plan is followed 
in some European cities of lighting all the lights every night from 
dusk to midnight or 1 o'clock, and then extinguishing part of 
them, sometimes every other light. Now and then this scheme 
is varied by having supplementary incandescent lamps attached 
to each arc pole and throwing these on during the morning hours. 
An arc pole thus arranged is shown in Fig. 155, and a beautiful 
example of similar practice using mantle burners as auxiliaries 
appears in Fig. 156. On the whole, this plan is likely to give 
better illumination than any form of moonlight schedule, but is 
less easily applicable here than abroad, since here most lights are 
on series circuits, while there the use of multiple connection is 
almost universal. No really effective scheme for cutting down 
the hours of lighting while yet adequately lighting the streets 
through the hours of darkness is reasonably to be expected. The 
only question that may fairly be raised is whether it may not 
be proper, say after midnight, on account of the changed condi- 
tions in the streets, to regard certain first-class streets as second- 
class streets, and hence to reduce the illumination in them by 


cutting out every other light, which, in a liberally lighted etreet, if 
planned for in advance, is not impracticable. The same reasoning 
might apply to a few second-class streets. A consistent application 
of this principle might reduce the average hours of lighting per 
year from the 4000 of the all-night schedule to some point between 
3000 and 3500 hours, depending on the number of lights affected. 
If rigorous economy in street lighting is absolutely necessary, this 
line is the logical one to follow. 

Fig. 155. 

Before passing to the practical design of street lighting, it is 
worth noting that while we have here reckoned the illumination 
as for normal incidence it is the usual practice abroad to reckon 
the horizontal component. This, as has already been seen, makes 
a very great difference in reckoning back from the required mini- 
mum illumination to the necessary power of the radiant. On the 
other hand, when reckoning the horizontal component one is at 
liberty to sum up the light received from both directions on a 
street or from all directions in an open space. The author per- 


sonally prefers to consider only normal incidence and lights from 
one direction only, since it is this condition which must be con- 
sidered in those uses of street lights which require the strongest 
illumination. If the lighting meets the severest requirement, it 
will also meet all the others. Objects of which the details are to 
be made out are generally held so as to be lighted from only one 
direction, and hence it is this which must be considered. In point 
of fact, the European practice is perfectly sound as regards the 

Kg. 156. 

results, because with a minimum requirement set quite as high 
as here indicated there is no doubt about getting sufficient nor- 
mal illumination when the horizontal requirements are fulfilled. 
Furthermore, with the radiants commonly employed for street 
lighting and spaced so as to get the required horizontal com- 
ponent, the height of the sources above the street is such as to 
approximately fulfill also the requirement for normal illumination. 
For instance, the common spacing for arc lights in Continental 
cities is for chief streets about 30 to 40 meters, and the lamps 


themselves being commonly hung 8 or 10 meters high, the angle 
of depression reckoned to the midway point rises to the vicinity of 
30 degrees, and hence it is numerically a matter of indifference 
whether one reckons the light received at this incidence normally 
from a single lamp or the horizontal component from a lamp on 
each side. Thus, in effect, the two methods of measurement lead 
to practically the same result when the lighting fulfills the neces- 
sary requirement for intensity on either theory of procedure. The 
plane of illumination, that at which the required intensity should 
be found, is commonly taken at 1 to 1.5 meters above the pave- 
ment merely as a matter of convenience, it being very difficult 
to use a photometer near the pavement in the street on account 
of traffic. The vertical component of illumination has practically 
seldom to be considered in street lighting, save in its effect on 
near-by buildings, a matter which will be taken up presently. 

Reviewing the nature of the problem of street lighting, that is, 
the illumination of a narrow surface extending in both directions 
from the source of light, it is obvious that the vertical distribu- 
tion of light around the radius is a matter of great importance. 
A uniform spherical distribution is bad for the purpose, and the 
practical question regarding an illuminant for such use is how 
much of its effective flux of light can be conveniently turned 
downward upon the street. Light above the horizontal is not 
absolutely wasted, for it does some service by illuminating build- 
ings. A radiant for street lighting should, however, be judged 
substantially by the lower hemispherical intensity, taking the 
lamp and its reflecting system together. Reflectors are useful 
with all varieties of street lamps merely for the purpose of de- 
flecting downward light which would be otherwise lost toward 
the sky, and within limits the natural distribution from the source, 
that is, the distribution without any reflector, is a matter of no 
great importance, save as it may influence the convenient design 
of the reflecting system. A radiant which naturally and without 
a reflector casts a large proportion of its light into the downward 
hemisphere does not gain materially by that peculiarity unless it 
can show greater efficiency in luminous flux per watt than some 
other lamp with its reflector, or possesses important practical quali- 
fications in its favor quite outside the matter of distribution. 

Praiseworthy efforts have been made toward securing by re- 
flectors a distribution stretching up and down the street rather 


than radially in all directions. They have, unfortunately, made 
little headway as yet for reasons psychological rather than phys- 
ical. Such reflectors are apt to be awkward in appearance, which 
is a great disadvantage by day, and require the maintenance of 
rather exact adjustment in order to do the most efficient work by 
night. Obviously, one could not push this sort of redistribution 
too far lest he should get an illumination approximating that 
which might have been obtained by a pair of automobile head- 
lights facing up and down the street from a pole top and giving 
a capital light at a distance, but little near by. If, however, one 
could obtain a distribution which, instead of being circular, was 
an ellipse or similar figure having a major axis two or three times 
the minor, it would prove of great practical service in street light- 
ing, but the improvement must not be at the expense of clumsy 
appearance or require constant care in cleaning to keep up its 

The modification of distribution is important from the stand- 
point of securing the proper spacing and height of lights. A dis- 
tribution curve with its maximum 50 or 60 degrees below the 
horizontal is disadvantageous in that it compels a lamp to be 
placed high in order to bring its zone of maximum flux of light 
out toward the radius of minimum illumination indicated by the 
power of the radiant. On the other hand, a maximum within 
15 degrees of the horizontal is almost equally bad, since thten the 
most effective rays can generally be made to give the required 
minimum only from a lamp placed so low that considerations of 
avoiding glare make it undesirable. From a practical standpoint, 
a maximum somewhere between 15 and 30 degrees below the 
horizontal is most desirable, considering the available power and 
the permissible height of most commercial radiants. 

These considerations bring one at once to the question of the 
spacing and height of lights for street lighting, and with this is 
inextricably bound up the troublesome question of large versus 
small units. Considering the usual circular distributions, it is 
readily seen that, basing judgment upon the required minimum 
and average values along the street, small units have the advan- 
tage in the total intensity required to meet given conditions. 
This total intensity, assuming radiants of the same distribution 
curve and placed at the best height, as indicated by equation 
(2), page 285, varies apparently inversely with the square of the 



number of units assigned to cover a given length of street. In 
other words, to double the effective radius of action of a light, 
preserving these conditions of symmetry, requires four times the 
intensity, and so on. A glance at Fig. 157 shows the condition of 
things: a is the radius of action for a given mimimn m for the 
source A, and b the similar radius for the symmetrically positioned 
radiant ft; clearly, if a second small radiant ft be placed so that 
its radius of action touches that of its mate, one obtains from 
either disposition the same minimum illumination along the center 
line of the street, but the two radiants ft and ft need each give 
only one-fourth the light given by A. Moreover, since the circle 

Fig. 157. 

of radius A has four times the area of one of the circles of radius 
B, the total flux per square meter is the same in either case if 
symmetry is preserved, and the average flux through the respective 
cones of distribution would be equal. If this rudimentary com- 
putation were all there were to the matter, the case would be 
definitely settled in favor of small lights, but it is easy to see 
that this would lead to a reduclio ad absurdum. For a single 
candle will give 0.2 meter-candle at a distance of nearly two and 
a quarter meters, and one may easily imagine the general darkness 
of a sidewalk, for instance, illuminated by candles nearly 4 j meters 
apart. The secret of the matter is, of course, the great and useful 
flux of light required to give everywhere the fixed minimum inten- 
sity when using powerful illuminants. 


In Fig. 157 if the light A, with its effective radius a, illuminates 
the circle of which it is the center, then the lights JSi and ft equally 
illuminate their respective circles, but there is an equal shaded 
area C shown outside them which they do not light, and which is 
effectively lighted by the radiant A. Now, this large area is, from 
the standpoint of street lighting, not useless. If the houses stand 
well back from the street, it effectively affords them police protec- 
tion; if they face close upon the street, some of the extra light is 
reflected from their surfaces and again becomes useful either as 
lighting directly the street or as furnishing a bright background 
against which dark objects may readily be seen. In other words 
while the minimum requirement, and even the average require- 
ment, can be met by a much smaller total flux with small units 
than with large ones, the latter do in fact add greatly to the effec- 
tiveness of lighting for the purposes of its use. 

Still further, if we replace one large light by n small lights under 
the assumed conditions, we not only get but one n ih the total flux 
for the same minimum, but we have to install and maintain n lights 
instead of one; and, since the investment and maintenance charges 
make up a large proportion of the total cost of any street illumi- 
nant, it often turns out that in an attempt to gain illuminating 
efficiency by decreasing the spacing and using small lamps there is 
no reduction in total cost commensurate with the loss of light flux. 
It is on this practical condition that the choice between large and 
small units usually depends. Also, if a certain minimum illumi- 
nation be set, and a source of large power be replaced by smaller 
sources, these cannot be placed as B Xt B 2 , B Zf B Af Fig. 158, with their 
circles of limit illumination merely tangent, for that leaves much 
of the street below the flowed minimum. They would have to be 
placed overlapping, like B b , JS 6 , B 7 , jB 8 , or else the power of each 
smaller radiant must be considerably increased. Without going 
into the analysis of the requirement of overlapping enough to cover 
the street at the required minimum, it is apparent that the economy 
in flux secured by using small sources is much smaller than at first 
seems plausible, and the gain in cost smaller still. In point of 
fact, small sources are advantageous chiefly in second-class and 
third-class lighting. For first-class lighting they will rarely be 
found economical. 

In thickly built-up and important streets the enormous light 
flux from powerful light sources is so useful for the general pur- 



poses of illumination that it pays to utilize it. Particularly is this 
the case since it is true both in gas and electric lighting that the 
large units are of very much higher efficiency than the small ones. 
On the other hand, in streets requiring only moderate illumination, 
and particularly in streets low-hung with shadowing trees, the small 
light, which for efficient use must be hung rather low, as we have 
seen, is greatly to be preferred. In streets of the third class, where 
the conditions are such that there are no lateral reflecting surfaces 
to be utilized, the small unit is imperative. In everyday practice, 
especially with electric illuminants, the distribution curves of the 
small incandescent units and the much larger arc units, even when 

Fig. 158. 

both are modified with the best available reflectors, vary very 
materially, so that in practice it is desirable to draw curves like 
L» (Fig. 149), giving illumination as a function of distance according 
to the lamp, and thus make graphic comparison of the relative 
results to be obtained in service. When installed in diffusing globes 
both arc and incandescent lamps, and gas lamps as well, tend to 
a rounded type of distribution, which makes reflectors necessary to 
secure maximum efficiency. As a rule, all illuminants in American 
practice are mounted lower than they ought to be for efficiency. 
Powerful arcs or equivalent gas lamps should generally be mounted 
at least 25 or 30 feet above the pavement, and in case of the very 
large units even considerably more, up to 50 or 60 feet. Lamps of 
the type of the larger incandescent electric units and the corre- 


sponding gas lights need to be carried to the vicinity of 15 feet 
high for economical results, varying somewhat with the type of 
reflectors employed. Practically, at the present time, one has to 
choose between radiants giving, say, 50 to 100 candle power, on 
one hand, or between 500 and 1000 on the other, only a few com- 
mercial sources running to still larger powers. This choice once 

Fig. 159. 

made, the considerations already given show the spacing which 
must be maintained in order to give the minimum illumination and 
the average illumination, respectively, suitable for the various 
classes of work, and the characteristics of the lamps used hold this 
spacing within narrower limits than seem at first sight probable. 

In much first-class lighting extra light received from shop win- 
dows and signs affords valuable reinforcement during the hours 
when most light is needed. This illumination is very commonly as 


great as that from the street lamps, and sometimes several times 
greater. It is useful, but one cannot safely count much upon it, 
since it is largely influenced by habit. 

In placing lamps, large or small, it is imperative that they should 
be so located that their useful light flux can be utilized. This con- 
dition is often violated by placing lamps where their light is very 
largely cut off by trees. By far the best method of placing street 
lamps from the standpoint of illumination is the cross-suspension. 
The best form of this, very general in Continental cities, is, in closely 

Rg. 160. 

built streets, the cross-suspension from buildings, well shown in 
Fig. 159. In this country there is seldom proper provision for this, 
so that one is driven to the clumsier suspension between poles. A 
street adequately lighted by lamps upon cross-suspensions may 
often fail of it when side poles are used. Nevertheless, owing to 
our local conditions, it is more general to use side poles. These 
are too often ugly in design and hence offensive by daylight. 
There is no excuse for this, save petty economy, since poles of 
graceful design are readily obtainable. A capital example of Con- 
tinental practice in side poles is shown in Fig. 160, a light design 


of steel tube springing from a cast-iron base, and bearing a flame 
arc. If powerful lights are used in streets which must be equipped 
with side poles, long mast arms, ugly as they are, are practically 
extremely useful. In fairly open streets, lamps bracketed out from 
2 to 6 or 8 feet from the curb give satisfactory results, and these 
methods of suspension are available for all illuminants, electric or 
other. Abroad these fixtures are often systematically placed on 
buildings whenever possible, thus relieving the street of poles. A 
good example of such brackets appears in Fig. 161 applied both to 

Kg. 161. 

an arc and to a gas lamp. Where there are many trees small units 
bracketed fairly well out from the curb are by far the most suc- 
cessful illuminants, although there may be local reasons for prefer- 
ring larger ones in cases where rather brilliant lighting is necessary, 
and there is practical or asthetic objection to increase in the num- 
ber of posts. In final warning in the matter of spacing and height, 
it must be said that only a few of the most powerful sources, as yet 
very little used in this country, are capable of giving the illumina- 
tion required for first-class streets, over a radius of as much as 150 
feet. Ordinary arcs spaced at 400 to 500 feet, as is too often the 
case, are rarely adequate even for second-class lighting, on account 


of the long and very faintly lighted spaces which must exist. Only 
the most powerful radiants may be thus spaced. 

The proper placing and screening of lights to avoid glare is 
another important matter. Glare is due to a number of causes, 
but practically it is chargeable to the use of sources of too great 
intrinsic brilliancy, or too great absolute intensity at short dis- 
tances. A powerful light, even at moderate intrinsic brilliancy, 
when viewed at short range floods the eye with light to an extent 
that interferes seriously with vision. It also cuts down the pupillary 
aperture to half or one-third of its normal value, which greatly 
diminishes the visibility of the less brilliantly illuminated part 
of the field, and, more than anything else, it spoils the dark- 
adaptation which makes enormously greater difference than any- 
thing due to pupillary reaction. At considerable distances there 
is very little trouble due to intrinsic brilliancy. As, however, one 
is constantly coming into close range with street lights, protection 
against too high brilliancy is imperative in case of powerful 
radiants, either gas or electric. The smaller lights, sending much 
less luminous energy to the eye, produce less disturbance by their 
glare, and diffusion, while desirable, is less necessary, save when 
one falls to third-class lighting, in which dark-adaptation is all- 
important. No trouble would be experienced with any ordinary 
illuminants if screened behind mildly diffusing globes. Unless 
lights are so screened the minimum illumination must be raised 
very materially for the same ease of seeing. 

Owing to the comparatively weak illumination in most street 
lighting, methods of measuring it are somewhat troublesome. 
There is always difficulty in photometry with very weak light on 
the photometer screen, and this is aggravated in street work by 
frequent unsteadiness of the lights, and in some cases by their 
great difference in color from the lights used as comparison stand- 
ards in field work. Comparisons of illumination near and below 
the minimum specified for second-class lighting are peculiarly falla- 
cious, owing to the disturbing effect of varying adaptation. Indeed, 
it is not putting it too strongly to say that comparisons of such kind, 
say at 0.2 or 0.3 meter-candle and below, are unreliable, even when 
made with the best available field photometers. Consistent results 
may sometimes be obtained by a single observer, or by two ob- 
servers so used to working together that their results are in no 
wise independent, but consistency is no proof of reliability. 


Acuity photometers sometimes used for such cases are even 
worse, and their results are not worthy of serious consideration 
as expressions of anything more than individual opinion. These 
instruments violate a fundamental rule of physical measurements, 
in that the quantity sought varies enormously with the one actually 
used for measurement. By all means the most reliable method of 
determining illumination is computing it from the known distri- 
bution curves of the radiants, taken not with carefully cleaned 
and adjusted lamps under laboratory conditions, but with lamps 
in their ordinary service condition run from the commercial wires 
or gas mains, even though tested in the laboratory. The effect 
of dirt on the inclosing globes is so serious that it must be taken 
into account in this way. 

For similar reasons the intercomparison in the field of different 
street illuminants is very unsatisfactory. One can tell photomet- 
rically pretty nearly what a light is doing, and can judge in a general 
way of the effectiveness of that particular lamp. He cannot form, 
however, a correct judgment of the relative performance of two 
lamps of different kinds unless they are conspicuously different 
from each other, since he does not know ordinarily whether each 
of the lights is burning under its normal conditions, whether one 
of them is ill adjusted, and with a globe considerably dirtier than 
the average, or whether the other has been carefully adjusted to 
give more than its normal duty, and is as clean as care can make 
it. Of course, anyone can tell a clean globe from a dirty globe, 
but he can do nothing more than guess how much difference to 
ascribe to this cause. 

One meets a good many cases of skillful jockeying with lamps, 
and even with the photometry of lamps, and while field com- 
parisons may be interesting as experiments, they do not form a 
suitable basis on which to found lighting contracts which may 
involve hundreds of thousands of dollars during their terms. And 
particularly is this stricture directed at reading tests so called, at 
low intensity, which are largely matters of adaptation and adroit 
manipulation of the conditions. It is well within bounds to say 
that they are generally open to suspicion, whether directed by the 
contractor for illumination or by critics inclined to be captious. 
They are sometimes called " practical," but experience teaches one 
to define a " practical " test as a test cunningly devised to divert 
attention from the objectionable points of the thing tested. Their 


proper sphere is merely to furnish one item of information, and 
not a very important one, about what the lights are doing. Two 
lights, both in good condition and adjusted and operated by im- 
partial observers, can be photometrically compared in the field 
with a reasonable degree of precision, but never any more satis- 
factorily than they can be compared under properly arranged 
laboratory conditions. 

To summarize briefly the characteristics of modern illuminants 
for street service, one may divide them into five generally familiar 
types: 1, flame or luminous arcs; 2, carbon arcs; 3, tungsten incan- 
descent lamps; 4, high-pressure mantle gas lamps; 5, low-pressure 
mantle gas lamps. As class 4 is little used in this country, one 
need say no more here than that the high-pressure gas lamps are 
powerful illuminants comparable in intensity to the flame arcs, 
that is, running from, say, 1000 to 3000 candle power, and of 
sufficiently good color and steadiness to meet all street require- 
ments. The low-pressure mantle gas lamps are usually of 100 
candle power or less, like tungsten incandescent lamps, and when 
properly maintained bear the same relation to the high-pressure 
lamps that the incandescent lamps do to the arcs. Considered 
merely as radiants for the street, they have no peculiarities of color 
or distribution which separate them from other radiants. 

To summarize the arc situation already treated in Chapter VIII, 
the flaming or luminous arcs are of three general classes: 1, arcs 
burning carbons of which one or both are mineralized, commonly 
with calcium fluoride for yellow light or with other substances for 
a whiter light, with converging carbons pointed in an acute angle 
downward; 2, vertical carbon flame arcs, commonly known as the 
system Blondel lamps, burning similar carbons heavily mineralized 
and in vertical position; 3, lamps burning electrodes, at least one 
of which is charged with metallic oxides, most commonly oxides 
of iron and titanium in various proportions. Such are the mag- 
netite arcs in common use. In this case the positive electrode is 
of copper and the active mineralized one is an iron tube packed 
with the oxides. All these lamps may run to high powers, from 
1000 to 2000 or 3000 candle power, in the effective zones of 
the lower hemisphere, and with mean lower hemispherical candle 
power ranging up to 2000 or more. The converging carbon lamps 
from the position of the electrodes tend to throw the light down- 
ward, to an extent that is not readily corrected by reflectors. 


They should hence be placed specially high with respect to the 
spacing. They are reasonably steady and have proved very sat- 
isfactory illuminants. The vertical carbon flame arcs give a dis- 
tribution which lends itself rather more readily to the successful 
use of reflectors, and the maximum light falls, with properly de- 
signed reflectors, within the useful zones from 15 to 30 degrees 
below the horizontal. They are, on the whole, more efficient for 
street service than the converging carbon lamps, and they are 
made even up to 3000 candle power and more. The lamp 
described with reference to the lighting of Copley Square is 
one of this class, and gives an intensity of about 2500 candle 
power, including the opal globe, in average condition, at the angle 
of about 15 degrees below the horizontal. This advantageous 
distribution is due largely to a well-placed reflector. The specific 
consumption of the large flame arcs of both varieties is some- 
where in the region of about one-quarter to one-third watt per 
candle. The metallic oxide lamps have for their chief advantage 
a rather long-burning electrode, giving within a reasonable length 
of pencil of 8 or 10 inches a life of 50 to 150 hours. The products 
of combustion, being brownish oxides, tend to smut the globes, 
and have to be gotten rid of by special draft channels which carry 
the fumes away. The commonest of this type is the so-called 
magnetite lamp, which has proved extremely successful as an illu- 
minant in practice. Its specific consumption, when put in a light- 
opal globe, ranges in the region between one-half watt and 1 watt 
per candle, according to current, as in other arc lights. It fur- 
nishes a suitably steady light in the vicinity of 1000 hemispherical 
candle power in the moderate sizes, and nearly as much again as 
an extreme figure. The color is good and the steadiness adequate, 
and the lamp has been rapidly driving out the older forms of arc, 
being preferred to other flaming arcs in this country on account 
of the long life of the electrodes. The magnetite lamp is operated 
at from 4 to 6 or 7 or even 10 amperes, the latter rarely, and the 
voltage at the arc is about 80. These characteristics make it very 
convenient for use on series circuits. 

Ordinary carbon arcs are becoming obsolete for street service. 
They run in sizes from 600 or 800 mean hemispherical candle 
power to as low as 200 or 250. The former figure belongs to 
the few powerful open arcs that are in existence, the latter to 
some of the alternating-current inclosed arcs. The specific con- 


sumptions range from a little better than 1 watt per candle in 
the former case to above 2 watts per candle in the latter. 

The tungsten incandescent lamps are too well known to need 
further comment here. They are ordinarily available in candle 
powers from 40 to 100, and lamps of 200 or 300 or even 400 candle 
power have been produced, but have not come into much use 
up to the present time. The tungsten lamps form the main reli- 
ance of street electric incandescent lighting at the present time. 
Their specific consumption is in the vicinity of 1J watts per 
candle, and their distribution, when suitably equipped with re- 
flectors, well suits street lighting. The carbon incandescent lamp, 
like the open-flame gas lamp, is rapidly becoming obsolete. 

Contracts for street lighting are essentially contracts for ser- 
vice on the part of a public supply corporation. A city does not 
buy merely a given number of kilowatt hours per annum, nor a 
specified number of arc or incandescent lamps of certain candle 
power. It does buy, in fact, light and service with specified illu- 
minants, including current or gas, as the case may be, maintenance 
of the lights in first-class working condition, and the operation of 
them for certain specified hours per year. It is the character of 
the service that determines the difference between good and bad 
lighting. One may specify a certain consumption of gas or of 
watts in a lamp and still get extremely bad service. He may also 
specify a certain minimum illumination and get extremely bad 
service. If he tries to buy illumination as such, he faces the 
practical difficulty of measuring it with sufficient precision ^f or the 
maintenance of contractual relations. He can tell with the pho- 
tometer whether the lights are performing well or badly, but he 
cannot by any means estimate the faint illumination customarily 
used in this country as a minimum with a degree of precision that 
should pass any conscientious auditing department. Buying and 
selling illumination as such is simply courting litigation. 

The soundest basis for a contract between a supply company 
and a municipality, for street lighting, is for service during speci- 
fied hours per year, and with proper allowances for "outages," of 
specified types of lamp, the characteristics of which can be evalu- 
ated, such lamps being placed in accordance with the requirements 
of the city. If they are placed so as to meet such requirements 
of illumination as have been previously set forth, and are properly 
maintained by the operating company, the illumination will be 


found adequate, and its value can be on the average reckoned 
from the known characteristics of the lamps with far greater pre- 
cision than it can be measured on the ground. The location of 
the lamps should be done under the direction of the municipality 
so as to produce the illumination required, but if any definition of 
the illumination is specified, both the minimum and the average 
should be included. Lighting on the basis of a contractual mini- 
mum only is certain to result in bad lighting. Judging by average 
illumination alone is merely an incentive to unequal distribution, 
but when the illuminants themselves and the terms of their output 
and operation are properly specified the illumination will take care 
of itself. 



There is a certain transitional region between street and other 
exterior lighting of a purely utilitarian character, and the illumi- 
nation in which the decorative element is predominant. In the 
lighting of public places, as has already been pointed out, the 
lighting fixtures and the distribution of the illumination should . 
have relation to the decorative possibilities of the place. This 

condition is realized particularly in the lighting of semi- or wholly- 
architectural things, like bridges and esplanades, the entrances and 
courtyards of public buildings, terraces, ornamental bits of park- 
way, and the like. In such places not only must the illumination 
lie harmonious and without glare, but the fixtures themselves by 
day and by night should be appropriate and decorative. Usually 
they violate all the canons of science and good taste, the fixtures 
having been picked out of the catalogue of some persistently 
intrusive salesman by a committee of politicians. 


The lighting fixtures in places where their purpose is essentially 
decorative require a fine artistic instinct to secure proper design, 
and close supervision to secure illuminating efficiency, and these 
two are seldom found together. As an example of the good effects 
which may be secured by properly designed lighting fixtures, the 
bit of the Thames Embankment, shown in Fig. 162, is one of the 
best that has come under the author's observation. The posts 
are highly decorative in the modeling and are surmounted by opal 
balls fitted with tungsten lamps which make the standards beauti- 
fully effective at night, casting an ample mellow light along the way. 
A second fine example of lighting under similar circumstances is 
shown in Fig. 163, — the Quai de Mt. Blanc in Geneva. Here 

Fig. 163. 

again the tall and ornate fixtures are given a decorative motive 
which harmonizes exceedingly well with the location and environ- 
ment, and the effect by night is altogether beautiful. The funda- 
mental principle in all such lighting is that the fixtures must form 
a suitable part of their environment, and be designed as objects of 
art and not as samples of ironmongery. 

In most instances the larger sizes of tungsten lamps are the 
best sources of light for such cases, though now and then are lamps 
may be used, and in case of need the larger mantle burners will 
give a good account of themselves. It is not the particular kind 
of light which counts in this class of work so much as adequate 
amount and steadiness combined with suitable fixtures. About 
the entrances of fine buildings, public or private, suitable lighting 


is less rare than in parks and other public places, for the simple 
reason that such matters are usually in the charge of the archi- 
tects who are not without artistic instinct. Little need here 
be said regarding fixtures except to emphasize the fact that 
elaboration in design is not desirable unless a rather large ap- 
propriation can be made to cover the expense. If fixtures must 
be had at moderate cost, they must be simple, and probably will 
be none the less artistic for this qualification. As an example of 
simple and harmonious treatment of even so utilitarian a source 
of light as an arc lamp, note Fig. 164 from the entrance to a 
German theater. One need not, however, mul- 
tiply instances of this sort; it is only necessary 
to impress the fact that extreme elaboration is 
not necessary to obtaining the desired results, 
but that if elaboration be attempted it must 
be carried through consistently without trying 
to shirk expense. 

To pass to the next phase of the subject, we 
must consider the lighting of structures, that is, 
the branch of illumination which is intended for 
decorative purposes to bring into prominence by 
night, buildings and monuments which are of 
artistic value by day. 

It is never a work of necessity, although often 
desirable as a suitable appreciation of public 
structures which are in themselves worth seeing. 
Its laws, are, therefore, rather those of aesthetics 
than those of engineering, albeit the engineering requires peculiar 
adroitness in order not to defeat the aesthetic end sought. No 
class of lighting is, upon the whole, worse done, and the few 
masters of it, like the late Mr. Stieringer, have excelled rather 
by instinctive genius than by the application of the precedents of 

As to the character of such lighting, it varies very widely, from 
the mere emphasis of salient details, or strong accentuation of 
particular objects, to the securing of startling scenic effects by 
flooding the surface with light or marking it out in lines of fire. 

There are, indeed, two distinct classes of structural lighting, one 
bringing forcefully out, as far as may be, the daylight values of 
the object; the other, being the evolution, with the structure 

Pig. 104. 


as a basis, of artistic results in no wise akin to the effects of 
daylight. An example of the former case is the lighting of a 
monument, or of the fagade of a building, so as to secure the full 
artistic value of the structure. Into the latter class falls naturally 
the display lighting of expositions and of special buildings and 
grounds, enfUe. 

The methods of illuminating structures are as various as the 
purposes for which the illumination is used. They may include 
merely the skillful application of ordinary illuminants, the enforce- 
ment of their effect by reflectors and searchlights, and the employ- 
ment of small lights in an infinite variety of ways. The former 
methods find their largest application in fighting monuments and 
facades; the latter, in the production of scenic effects, in which 
the more powerful illuminants also may be made to play a most 
useful part. 

Broadly, one may divide the classes of effects sought in lighting 
structures into those which have to do with the lighting of surfaces 
as a whole, those in which particular portions of surfaces are sought 
to be emphasized, and those which are essentially scenic and 
decorative in their effects and bring into prominence not surfaces 
but outlines. Each kind has its legitimate field, but its applica- 
bility depends in each case on quite different criteria. In general, 
the first two classes belong essentially to structures beautiful in 
themselves, while the last named, if skillfully carried out, which 
it generally is not, may lend distinction to things comparatively 

The superficial lighting of structures, as of the whole fagade of a 
great building, is both difficult to do well and somewhat expensive. 
Particularly does its apparent expense run high, inasmuch as it 
is a case of deliberately pouring a flood of light on an exterior, 
generally from a source quite outside the building. The sources 
of light have a certain detached character that brings their cost 
sharply into view, while the same expense applied to even ineffi- 
cient and inartistic grouping of small lights about the structure 
itself would fail to produce the same psychological effect on the 
auditing department. 


In order to be effective, surface lighting must be both somewhat 
brilliant and very carefully directed. The greatest difficulty in 
getting a satisfactory result is that due to obtaining the proper 
direction of illumination. The fagade of a building ordinarily is 


lighted obliquely from above, without sharp shadows, save when 
the building is in brilliant sunshine. The strength of the illumina- 
tion falling on a building by daylight may easily run to many 
hundred or even thousand meter-candles. It is enough, at all 
events, to bring out a wealth of fine detail even in a very dark 

The coefficient of reflection of a building surface is usually rather 
small, say from 10 to 25 per cent, at ordinary angles of view, rising 
notably higher than this, say to between 30 and 40 per cent, only 
in case of buildings of very light and clean brick or stone, or of 
very lightly tinted concrete. In .most cities the accumulation 
of dirt due to smoke keeps the reflecting powGr low. Hence it 
takes strong lighting, such as is to be had by daylight, to bring the 
architectural details of a building out at anything like their full 
value, if they are of any delicacy. In the artificial lighting of a 
facade, the direction of the illumination is necessarily rather from 
below than from above, and unless the illumination is deliberately 
planned to provide a dominant direction of lighting the effect is 
usually to flatten out the projections and sink the detail into 
insignificance. Light coming indiscriminately from all azimuths 
along the front is likely to give a disagreeable shadowless effect, 
and the delicacy of the surface of the structure is quite lost. The 
illumination should, therefore, be given a predominant direction 
so as not to lose the effect of light and shade, and, in fact, some- 
what to exaggerate it in order to bring out something of the 
texture in a light dim compared with daylight. 

If we could bring ourselves to a really progressive frame of 
mind, searchlights and reflectors used from points well outside 
the building to be illuminated could be made to produce much 
better results than are obtained by any other means, and there 
is much to be said for so lighting a fagade that it shows its archi- 
tectural value as it is, and not with the addition of freakish lighting 
effects generally undignified and sometimes ludicrous. 

A facade with striking features, large and dignified, is compara- 
tively easy to illuminate, while one with a wealth of fine detail 
requires so much light that the feat of dealing with it adequately 
is almost impossible. 

Searchlights and reflectors form the best means of getting ade- 
quate surface lighting where large areas are concerned. A slight 
digression regarding the searchlight for this and similar purposes 


may be here permissible, since the properties of the searchlight 
as an illuminant are generally imperfectly understood. From the 
standpoint of luminous flux, the case of the searchlight is a com- 
paratively easy one. From the energy consumed in the arc and 
the structure of the lamp, it is not difficult to form an approximate 
idea of the lumens which emerge from the system. Then the illu- 
mination received on any surface is this total flux divided by the 
area of the surface in square feet, if one wishes to express it in 
foot-candles, or in square meters if he chooses, as is preferable, the 
meter-candle as the unit. 
The illumination delivered by a searchlight system is 

4 weirj 4 eirj 

L = 

tit 2 r 2 


where e is the voltage at the arc, i the current, and t/ is the specific 
efficiency in mean spherical candle power per watt X the net reflec- 
tive coefficient of the mirror system, r being the radius of the inci- 
dent beam. 

17 = 1 

e=80v. T 4X80X50X1 1AA . ,. n . 

. -„ L = tt^t = 160 meter-candles. (1) 

i = 50 amps. 100 v 

r = 10 

Or at the just assumed voltage and radius, 100 meter-candles would 


. Lr 2 100X100 Q1 , Q . 

1 ~ 4^ = 4X80X1 = 3L2 "^ (2) 

Or to obtain tj, 

If L = 100 

* = 80 _Lr* _ 100X100 _ n7fi ^ 

r - 10 * ~~ Ul " 4 X 80 X 40 " °' 78 - {3) 

i =40 

rj expresses the specific efficiency of the searchlight system as a 
whole, and should be the subject of systematic experiments. 

This rule holds for cases in which the cosine of the semiangular 
aperture of the beam is near unity, i.e., when the measured illu- 
mination is substantially normal; (1) and (2) are subject to a simi- 
lar limitation. As to the absorption by the atmosphere, it is in 
clear weather small, amounting, from Sir William Abney's data, 


to about 15 per cent for the chief luminous rays in transmission 
through the whole thickness of the atmosphere. 

The flux measured in lumens remains the same, barring absorp- 
tion by the atmosphere, at all distances from the light, and the 
intensity at the surface illuminated becomes merely a matter of 
the area of the beam. 

Without the use of lights comparatively distant from the surface 
to be illuminated, surface lighting becomes increasingly difficult. 
It can be carried out to a certain extent with lights on the struc- 
ture itself, but the effect is not good if more than special portions 
of the surface are so illuminated. It is an extremely difficult matter 
to light a surface adequately from a point near itself without 
either making the light sources too conspicuous or rendering the 
illumination very uneven. Moreover, lighting at nearly grazing 
incidence distorts all surface details and destroys their delicacy. 
It may even produce extremely bizarre and unpleasant effects. 
Striking examples of the failure of illumination at grazing incidence 
may be found in the case of attempts to light paintings from reflec- 
tors placed near the plane of the canvas, the effect of which is to 
bring into glaring prominence every brush mark, quite destroying 
the effect the artist intended to produce. 

Only in rare instances can the lighting of a building by sources 
placed upon it prove effective, and then only when comparatively 
limited areas are sought to be illuminated, or when the effect 
intended to be produced is not that of daylight illumination but 
that of a special form of decoration. A striking success in this line 
was the lighting of the Metropolitan Life tower in New York during 
the Hudson-Fulton celebration, in which advantage was taken of 
the structure of the higher parts of the tower so to place the lights 
upon it as to bring the massive detail high in air into brilliant 
prominence against the sky. Fig. 165 gives a rather inadequate 
view of the conspicuously good result. 

Spot lighting is, except in such instances as that just mentioned, 
generally confined to the illumination of monuments or groups of 
statuary. It too frequently would be better to leave these to the 
kindly concealment of night, but now and then they are worth 
the effort at illumination. As a rule, attempts to light such things 
fail from placing the lights too near, and thereby producing dis- 
torted shadows which quite destroy the artistic value sought. Only 
massive, plain surfaces, such as the Washington Monument in 


;. 165 — Illumination of the Metropolitan Life Tower. 


Washington presents, can be adequately lighted from sources 
placed near the base. Monuments of ordinary types should be 
illuminated, if at all, from distances several times their height. 

Now and then on white surfaces, colored illumination can be 
used with beautiful effect; but these cases are rare and chiefly con- 
fined to temporary exposition work, in which there is small chance 
for time to destroy the high reflecting power of the surfaces neces- 
sary to brilliant effects. 

Eg. 166.— How not to do ft. 

Exposition lighting ia an art almost by itself, owing to the im- 
mense areas that have to be dealt with, and the extreme difficulty 
of getting suitable locations for lighting buildings from the outside. 
The main reliance in such work has in the past usually been out- 
lining by myriads of small incandescents. If skillfully done, that 
is, done with due reference to the magnitudes and distances of the 
buildings so as to preserve the ensemble effect by night, the result 
may be extremely beautiful ; but if the salient features are wrongly 


chosen or the ornamentation is exaggerated, nothing can, on the 
whole, return less of artistic value for the energy employed. At 
its worst, outlining becomes a mere symbolizing of structural lines, 
as a child might draw them upon a slate. Fig. 166, a night photo- 
graph of an important building, during the Hudson-Fulton cele- 
bration, will serve as a horrible example of how not to do it. 
Restraint and keen appreciation of the features of a building worth 
outlining are necessary to the securing of good results, otherwise 
the eye will simply be confused by a multitude of lights with nothing 
to indicate their distance, and will find even individual buildings 
distorted by the wrong spacing or placing of the lights. Fig. 167, 

Fig. 167. — The Electric Tower at Buffalo. 

the electric tower at the Pan-American Exposition of 1901, is a 
beautiful example of the harmonious application of correct prin- 
ciples in decorative illumination, a masterpiece in its way, due to 
the consummate skill of the late Mr. Stieringer. Skillfully used, 
outlining may confer by night singular beauty on structures either 
commonplace or grimly utilitarian by day. Very striking examples 
of the decorative use of outlining and similar illuminative devices 
were shown during the Hudson-Fulton celebration in New York. 
No one who saw the East . River bridges and the stacks of the 


Water-side station failed to recognize the great artistic value of 
judiciously strung lights. The bridges from a distance were things 
of glory instead of grimy skeletons of cable and girder, while the 
stacks, by day merely solid and purposeful, became beautifully 
decorated symbolic towers of light. A fine example of graceful 
outlining was furnished in the illumination of the Eiffel Tower at 
the Paris Exposition of 1900, shown in Fig. 168. 

In attempting to outline buildings, it is generally found best to 
emphasize some of the special features as well as the general con- 

Fig. 168, 

tours, so that the illumination is not only structural but decorative. 
In great measure success depends on the judgment of the engineer 
in fitting the spacing and power of his lights to the particular work 
in hand. Fig. 169 gives a night view of the Electricity Building 
at the St. Louis Exposition of 1904. It is an admirable type of 
combined outlining and decorative lighting, planned for a view- 
point across the lagoon. In such work it makes a very material 
difference whether the lighting as a whole is to be viewed from 
a distance or near by, from practically on a level or from below. 
As to the proper spacing of lights for general service in such work, 
it usually runs in practice from 8 inches to 2 feet, while in some 


instances even these dimensions may be passed. The fundamental 
thing is to proportion the spacing to the ordinary viewing distance, 
so that the lights will neither run together in a blurred line nor 
present a scattered appearance. 

The former limit in the last resort depends on the power of the 
eye to separate two neighboring luminous points. Many experi- 
ments on this sort of visual acuity have been made, and they may 
be summed up by saying that the eye distinguishes two bright 
points as separate very easily when they subtend a visual angle of 

Pig. 169. 

5 minutes of arc; fairly well when they subtend an angle of 3 
minutes; and under favorable circumstances, and with difficulty, 
when they subtend an angle of 1 minute. TheBe figures are some- 
what influenced by the actual darkness of the background, and by 
the actual intensity of the lights with respect to their production 
of irradiation. 

Now, an angle of 1 minute is subtended by two points distant 
from each other by 0.0003 of the viewing distance. The 3-minute 
angle, therefore, corresponds nearly to a separation of one part in 
a thousand, and the 5-minute angle to one part in six to seven 


hundred. Around about this latter figure good results are obtain- 
able, and the separation of lamps can often be carried up to 10 
minutes of arc with advantage. Closer spacing than 3 minutes is 
seldom desirable, since only in rare cases does one wish to produce 
the effect of continuous lines. There is, therefore, a wide range in 
spacing permissible with which to take account of the important 
questions relating to perspective. The frontispiece shows a remark- 
ably distinguished and successful use of festooned lights in the 
Court of Honor at the Hudson-Fulton celebration of 1909. Two 
blocks along Fifth Avenue, from 42nd Street to 40th Street, were 
included, and 36 massive pylons were erected along the avenue to 
bear the central decorations of the occasion. The night effect was 
very striking and beautiful, though the plate is marred by the 
streaks due to the headlights of motor cars. 

The effect of the spacing, intensity, and alignment of lights upon 
nocturnal perspective has played a very small part heretofore in 
illumination, although it is well understood as a practical art by 
the masters of stagecraft. Ordinarily, one wishes to preserve the 
normal conditions of perspective in undertaking artificial illumi- 
nation. This requirement implies a generally uniform spacing of 
lights, since the eye instinctively judges the length of a line of bright 
points by their apparent approximation as they reach the vanish- 
ing point. 

In the case of lines of light generally viewed obliquely, the spac- 
ing may, however, be widened, since the visual angle between 
points in such case corresponds to a narrower spacing than when 
viewed normally. For example, lines of lights festooned lengthwise 
of a street may be spread far more widely than usual, while still 
preserving unity of effect, since they are, upon the whole, viewed 
always from a very oblique angle. Lines of festoons thrown cross- 
wise of the same street are always seen normally to their length, 
and consequently should revert to standard spacing. 

There are, however, many instances in which lights can be ad- 
vantageously used, not to preserve the perspective, but to force it 
and to create illusions of distance. If one were so placed as to look 
down a long street, viewing it from a fixed point and not passing 
along it, it would be possible to produce extraordinary illusions of 
perspective by varying the spacing and the intensity of the lights. 
In the absence of any permanent objects upon which the eye can 
fall to determine distance, it is compelled to judge very largely by 


the apparent perspective. A row of lights down each side of the 
street, diminishing in spacing or intensity, or both, would infallibly 
call to the mind the conception of indefinite distance stretching out 
into the night. If, on the other hand, the spacing were progres- 
sively increased and the intensity also increased, each within limits, 
the effect would be to produce an apparent shortening of the vista. 
These effects may be very pronounced even where there is no 
deliberate angular shifting of lines in the field of vision to produce 
illusions of perspective. Lines converged toward an artificial vanish- 
ing point abnormally near are, of course, familiar in stagecraft, and 
by adjusting the stage setting on a diminishing scale, with lines thus 
converged, it is possible to create in great perfection the illusion of 
a far-reaching space, even on a stage of very modest dimensions. 

These effects, emphasized by powerful lighting in the near fore- 
ground, diminishing toward the rear, are quite familiar, and are 
often, in fact, overdone. When properly carried out they are 
immensely striking in effect. The forcing of perspective in this 
way and the taking advantage of the characteristics of vision to 
create illusions of direction and distance have been known at least 
since the time of the builders of the great monuments of Greece. 
Not only did these masters swell their columns slightly to overcome 
the illusion of outline presented to the eye at a relatively near view- 
point, but they even drew the columns together and toward the 
structure slightly at the top, as in the Parthenon, by an amount 
not large enough to be conspicuous, yet sufficient to accentuate the 
height. They knew well, too, how to proportion the scale of their 
ornamentation to the viewpoint, and seemed by a fine instinct to 
have discovered much that in practice has been too often forgotten 
in the centuries that have intervened. Now, this same sort of 
effect can be produced by judicious modification of the external 
lighting of structures. For example, the author has had occasion 
to overcome the tendency of a building, somewhat too tall for 
symmetry, to vanish into indefinite height as night came on, by 
powerfully emphasizing the decorative lighting of the cornice and 
spacing a horizontal line of lights across the front, disproportion- 
ately close compared with those above. The effect, as the lights 
come on in a dark evening, and the towering top emerges out of dis- 
tant blackness into its proper position, is somewhat striking. By 
such devices as these one can not only overcome the curious dis- 
tortions produced by night, but can, if necessary, create a wide 


variety of illusions on a large scale, less perfect, perhaps, but 
almost as striking, as those worked out upon the stage. 

The effect of the intensity of lights in such work is worth noting. 
It is a curious fact that the eye carries so very imperfect con- 
ceptions of intensity. For outlining work, signs and similar uses, 
8-candle-power lamps are practically as good as 16's, and 4's about 
as effective as either. The author has actually changed 8-candle- 
power lamps for 4's on one line of a sign, leaving the others 
unchanged, without producing any noticeable difference whatever 
when the sign was viewed from the ordinary distance of several 
hundred yards. Any brilliant spot, however small, seems to serve 
the purpose, and the size of the lamps used is really determined 
rather by the ability to procure them than by anything else. 

Recent progress in sign work has tended to smaller and smaller 
lights with positive gain in the effect produced. In attempting, 
therefore, to create illusions by changing the size of lights, the 
change has to be an exaggerated one. The use of colored lights 
in such cases as we have under consideration has been barely 
touched upon in practice. It is made immensely effective in signs, 
and has been used successfully in some exposition work for purely 
decorative purposes, but color as an element in scenic illusion off 
the stage has scarcely been tried. It possesses, nevertheless, pos- 
sibilities which are worth much more intelligent study than has 
yet been given them. 

Colored light can be effectively used with reflector arcs, on white 
surfaces, on cascades, in fountains, and the like, but is seldom 
successful when tried with incandescent lamps, save on a very 
small scale. The difficulty lies in the dimness of colored bulbs and 
the failure of attempts to get delicate tints in this way. Colored 
glass bulbs are expensive, and coated bulbs accumulate dust and 
are seldom weatherproof. 

Much decorative lighting is for temporary purposes, but with 
the present facilities for obtaining current and the temporary 
mountings that can readily be obtained, the work is comparatively 

Special receptacles for signs and decorative designs are now 
made in convenient form for quickly putting together, and enable 
temporary work for special occasions to be very easily done. 
Fig. 170 shows one useful form of mounting device, in which 
the weatherproof receptacles can be quickly strung together with 


clamps and held neatly spaced in any way desirable. For decora- 
tive work on a considerable scale the retaining clamps would, of 
course, be much longer than here shown. 

There is a fine chance for art in turning on the lights in archi- 
tectural and other decorative work. The water rheostat, bringing 
all the lights simultaneously from a dull-red glow to full brilliancy, 
is by far the most comprehensive scheme for the purpose. In the 
absence of this, or in permanent work of which only a part is 
regularly used, the circuits should be so arranged as to allow a 
perfectly symmetrical development of the lighting without throw- 
ing on a very large current at any one time. 

In any and all decorative work the illumination must be sub- 
ordinated to the general architectural effect. Sins against art in 
this respect are all too common. Imagine, for example, a Doric 
temple with arc lights at the corners of the roof and festoons of red, 

Fig. 170. — Chun of Receptacles. 

white, and blue incandescents hung between the columns. About 
a structure of such severe simplicity lights must be used with 
extreme caution, while more ornate buildings can be treated with 
far greater freedom of decoration. 

It requires both fine artistic instinct and great technical skill 
to cope adequately with the problems of decorative illumination. 
The tricks of the art are manifold, and mostly meretricious. The 
facility with which electric currents may be manipulated is a con- 
tinual temptation to indulge in the ingenious and the spectacular 
without due regard for the unity of the results. 

Another class of work, hardly a part of ordinary lighting, but 
yet of considerable interest, is the use of lights purely for decorative 
purposes in interiors, in halls and auditoriums for special designs 
and as part of the decorative scheme of ballrooms and the like. 
This is really a branch of the art due entirely to electric lighting 
— since only by this means can it be rendered fully serviceable. 


Most branches of illumination are in a measure independent of 
the particular radiants employed. But the ease and safety with 
which incandescent lamps can be installed render them peculiarly 
applicable to such interior work. 

In operating on a comparatively large scale, all sorts of decora- 
tive designs can be carried out by means of 4-c.p., 8-c.p. or l&-c.p. 
lamps strung together in receptacles, in the manner of Fig. 171, or 
otherwise temporarily mounted for the purpose. For work on a 
smaller scale, or in the preparation of very elaborate designs, other 
means may be employed. 

For purely decorative purposes the miniature lamps serve a very 
useful purpose. Regular incandescents are made down to 6, 4, 
or even 2 candle power, but, as has already been explained, the 
filaments for these powers at ordinary voltages must needs be 
very slender and fragile, and the lamps are somewhat bulky. 

Hence for many uses it is better to make miniature lamps for 
connection in series, each lamp taking 5 to 25 volts to bring it 
to normal candle power. Imagine a 16-c.p. 100- 
volt lamp filament cut into four equal parts, and 
each of these parts mounted in a separate small 
bulb, and you have a clear idea of the principle 
involved. Commonly the miniature lamps for 
circuits of 100 to 125 volts are of 5 or 6 candle 
power, and connected five or even ten in series 
across the ordinary lighting mains. Fig. 171 
gives an excellent idea of the size and appear- 
ance of the perfectly plain miniature lamp. It 
is fitted to a tiny socket of the same general 
construction as the standard sockets for ordinary 
. lamps, but taking up so little room that the 
t I ' ~nd """t ' am P 8 can conveniently be assembled in almost 
Lamp any desired form. 

It is not altogether easy to manufacture these 
lamps so as to attain the uniformity necessary, if the lamps are to 
be run in series, and this at present constitutes a serious obstacle 
to their use on a large scalei They are generally not of high 
efficiency, since great uniformity and good life are the qualities 
most important. 

They can be fitted with tiny ornamental shades, and may be 
obtained of various shapes and colors, so that very elaborate 



decorative designs can be built up of them. In indoor work 
colored lamps may be freely used, and are capable of producing 
some very beautiful effects, but the plain or ordinary frosted lamps 
are most generally used. 

Owing to the small size of the sockets and fittings, the miniature 
lamps can be packed so closely as to produce the effect of an 
almost uniform line of light at comparatively small distances, 
so that most ornate schemes of ornamental illumination can be 
carried out by their aid. They are also very useful in building 
up small illuminated signs. At present many small tungsten lamps 
with small bulbs in standard sockets are in use. They take com- 
monly 5 to 10 watts at 10 volts or so, and are supplied in multiple 
from special transformers if worked on alternating current, or in 
series if worked on direct current. 

Lamps of special sizes and shapes, from a tiny J-c.p. bulb, hardly 
bigger than a large pea, to the candle-shaped lamp of 5 or 6 candle 
power, are sometimes used with good effect in interior decoration. 
When a regular electrical supply is not available, these little lamps 
can be obtained for very moderate voltages, say, from 5 to 10 volts, 
and can be run in parallel from storage cells, or even from primary 
batteries, for temporary use. All these miniature lamps can now 
be had with tantalum or with tungsten filaments which greatly 
improve the situation, especially if one has to work from batteries. 

Such small lamps are sometimes used in the table decorations 
for banquets, and for kindred purposes. By their aid surprising 
and beautiful effects are attainable, which would be quite impos- 
sible with any flame illuminant. But they must be cautiously 
used, for their very facility tends to encourage their employment 
in effects more bizarre than artistic. 

It is well, too, to add a word of caution as regards the possible 
danger from fire. It is so easy to wire for incandescents that, par- 
ticularly when using miniature lamps, there is a natural tendency 
to rush the work at the expense of safety. Lamps in series on a 
110-volt circuit are quite capable of dangerous results if anything 
goes wrong, and even the battery lamps are not absolutely safe in 
the presence of inflammable material. 

It should, therefore, be an invariable rule not to install a tem- 
porary decorative circuit without the same attention to detail that 
would be exercised in a temporary circuit of the ordinary incan- 
descents. The same precautions are not always necessary, but all 


the wiring should be carefully done, joints should be fully protected, 
and, particularly, lamps should be kept out of contact with inflam- 
mable material. 

The incandescent lamp is often commended as producing little 
heat, and, in fact, as compared with other illuminants, its heating 
power is small. But a vessel of water can be boiled by plunging 
an ordinary 16-c.p. lamp in it nearly up to the socket, and cloth 
wrapped about such a lamp will infallibly be ignited within a com- 
paratively short time. The fact that the cloth does not burst into 
flame in a few minutes does not indicate safety, for time is an 
important element in ignition, and even an overheated steam pipe 
is capable of setting a fire, low as its temperature is. A good many 
fires have been started in shop windows by hanging fabrics too 
near to incandescent lamps, and even the miniature lamps are 
quite capable of similar mischief if in contact with anything easily 
inflamed. No illuminant has so high an efficiency that it produces 
a negligible amount of heat from the standpoint of fire risk. 

Special cable is now made to which lights can be attached with 
great facility, and by this means temporary work may be quickly 
and safely done. 

In ordinary domestic illumination miniature lamps have very 
little place. Nothing is to be saved by using them so long as they 
must be used in series at ordinary voltages. Now and then a 2- 
or 4-c.p. lamp may be useful as a night lamp, but it is better to 
use an ordinary lamp of moderate efficiency than to try miniature 
lamps. Sometimes, however, a circuit of miniature lamps may be 
installed for a dining room or a ballroom with excellent artistic 
results. In such cases it is better to use frosted than plain lamps, 
and, as a rule, colored lamps should be avoided, on account of 
the difficulty of getting delicate tints to show effectively. 

Temporary decorative circuits may, however, be very useful in 
domestic illumination for fetes and the like, in which case delicately 
colored ornamental shades can be applied or the lamps may be 
used in Japanese lanterns. Any country house fitted for electric 
lights can be temporarily wired for such purposes rather easily, and 
out-of-door temporary wiring can be installed without the rigid 
precautions necessary indoors. 

In all decorative lighting it is important to recognize the fact 
that illumination is a means to an artistic end, and not of itself the 
primary object. One is, in these days of electric lighting, far more 


likely to err by providing too much light than by failing to supply 

Great brilliancy is far less important than good distribution and 
freedom from glare. It is highly probable, for instance, that the 
effect of the illumination of the Electric Tower at the Pan-Ameri- 
can Exposition would have been seriously injured by the substitu- 
tion of 32-c.p. lamps for the 8-c.p. actually used, and it is absolutely 
certain that a dozen arc lights injudiciously placed would have 
detracted greatly from the harmonious result. 

In interior illumination the same rule holds true. By the reck- 
less use of brilliant radiants one can key the vision up to a point 
where its power of appreciating values in illumination is almost 
entirely lost. In decorative lighting great care must be used not 
to approach this point, but to leave the relief afforded by light 
and shade, and to realize the perspective in the details of the 

The commonest cause of failure in proper illumination is thrust- 
ing a brilliant light between the spectator and the object to be 
viewed, with the inevitable result of losing detail and hurting the 
eyes. Brilliant diffused light is in this particular only less objec- 
tionable than direct light, and both should be assiduously avoided. 

It must not be supposed that decorative lighting must necessarily 
be electric, since very beautiful results were attained before electric 
light was heard of, but electric lighting is unquestionably the most 
facile means of securing artistic results on a large scale. 




At the present time the ordinary materials of illumination are 
pretty well understood, and their proper use is a matter of good 
judgment and artistic sense. Illumination is not an exact science 
with well-defined laws of what one might call illuminative engineer- 
ing, but an art whereto an indefinable and incommunicable skill 
pertains almost as it does in the magic of the painter. 

There are certain general rules to be followed, certain utilitarian 
ends which must be reached at all hazards, but whether the result 
is brilliantly successful or hopelessly commonplace depends on the 
skill that inspires it. There must be in effective illumination a 
constant adaptation of means to ends, and a fine appreciation of 
values that quite defies description. One may attack the problem 
of illuminating a great building with all the resources of electrical 
engineering at his command, and score a garish failure, or he may 
conceivably be confined to the meager bounds of lamps and candles, 
and still triumph. 

The general tendency with the modern intense radiants at com- 
mand is to light too brilliantly, to key the vision to so high a pitch 
that it fails to appreciate the values of the chiaro-oscuro on which 
the artistic result depends. 

The desideratum in illumination, except for a small group of 
scenic effects, is the possession of cheap and fairly powerful 
radiants of low intrinsic brilliancy, capable of modification in 
delicate color tones. It is doubtful whether these qualities are 
compatible with very high luminous efficiency in a flame or incan- 
descent radiant. In modern gas and electric lighting the progress 
toward efficiency is in the direction of very high temperature, which 
implies high intrinsic brilliancy. 

Vacuum tube lamps give hope of better things, but at great 
risk of color difficulties, particularly if high efficiency is reached. 

Ideally, a gaseous radiant, with nearly its whole luminous energy 
concentrated in the visible spectrum, would give magnificent 
efficiency, but it by no means follows that it would give a good 



light. Sodium vapor meets the requirements just noted tolerably 
well, yet there is no more ghastly light than that given by a salted 
spirit lamp. 

It might be possible to work with a mixture of gases such 
as would give a light approximately white to the eye, and yet 
be very far from a practicable illuminant; for the phenomena of 
selective absorption are such, as we have already seen, that the 
color of a delicately tinted fabric depends on its receiving a certain 
scale of colors in the light which it reflects. To the eye a much 
simpler color scheme is competent to reproduce light substantially 
white, and such light falling on a colored fabric would by no means 
necessarily bring out the tints that glow by daylight. 

Even the firefly's secret, could man once penetrate it, might 
not prove such a valuable acquisition as it would seem at first 
thought. To the eye the light of most species seems greenish, 
and, in point of fact, it so completely lacks the full red and the 
violet rays that its effect as an illuminant on a large scale would be 
rather unpleasant, far worse than an early Welsbach at its most 
evil stage of decrepitude. We must not only steal the firefly's 
secret, but give him a few useful hints on the theory of color 
before the net result will be satisfactory from the aesthetic stand- 
point. Firefly light might do for a factory, but it would find but 
a poor market as a household illuminant. 

It is a somewhat difficult matter satisfactorily to define the 
'efficiency of an illuminant. Luminosity depends, like sound, upon 
the physiological relations of a certain form of energy, and cannot 
be directly reduced to a mechanical equivalent. 

The commonest conception of the efficiency of an illuminant 
is to regard it as defined by the proportion of the total radiant 
energy which is of luminous wave lengths. From this point of 
view the efficiency may approach unity either by the absence 
of infra-red and ultra-violet rays, — in other words, by purely 
selective radiation, — or by so great an increase of radiation in 
the visible spectrum as to render the energy of the remainder 
nearly negligible. 

In the former sense the luminous radiation of the firefly is of 
perfect efficiency; but, obviously, a purely monochromatic light 
utilizing the same total amount of energy might give a vastly 
better illumination — or a much worse one, according to the wave 
length of the light in relation to its effect on the eye. 


On the other hand, an intensive arc between tiny pencils of the 
material used for Nernst glowers is reputed to give, so far as watts 
per candle power go, an efficiency nearly as good as can be claimed 
for the firefly. The experiments in this case are perhaps not 
beyond cavil, but, even granting their substantial accuracy, it is 
perfectly certain that such an arc gives radiation by no means 
confined to the visible spectrum. 

The most that can be said in a definite way is that, assuming a 
continuous spectrum with its maximum luminous intensity in the 
yellow or yellowish green, there seems to be little chance of doing 
much better than about 0.2 watt per candle power. 

As a matter of fact, this efficiency is not approached by any 
practical illuminant giving a continuous spectrum. It has been 
reached and passed by some of the yellow-flame arcs burning 
carbons charged with calcium fluoride, of which the spectrum has 
its most intense bands in the region near to the highest point 
in the luminosity curve of the human eye. Of lights giving an 
approximately white light, the most efficient is the flame arc using 
carbons impregnated with ceria and some similar substances, 
by-products of the Welsbach industry, which closely approaches 
but does not quite equal the figures just given for the yellow- 
flaming arc. The white arc loses somewhat from the fact that it 
is white, and consequently to secure this color must contain rays 
of lower specific luminosity than those of the calcium-fluoride arc. 
The luminous arcs charged with iron and titanium can be pushed 
to somewhere between 0.5 and 0.75 watt per mean spherical candle 
power, and the most efficient of the open-carbon arcs may closely 
approach the latter figure. Carbon incandescent lamps scarcely 
do better than 3 to 4 watts per mean spherical candle power, and 
even the tungsten and other metallic filament lamps more recently 
introduced show a specific consumption not better than 1.5 watts 
per mean spherical candle power. 

Lamps employing incandescent gas or vapor vary over a con- 
siderable range, according to the spectral characteristics of the 
light and other properties of gas or vapor involved. The specific 
consumption of the intensive mercury arcs is approximately the 
same as that of the white flame arcs, that is, 0.25 to 0.3 watt 
per candle, the ordinary mercury arcs having a specific consumption 
of about twice this figure. Were all the energy concentrated in 
the green mercury line a startling improvement would be made, 


since the luminous efficiency actually found for this line exceeds 
50 candles per watt. 

The Moore tube, worked far less intensively than the mercury 
arcs, scarcely reaches a specific consumption of 2 watts per candle 
with the gases ordinarily available, while the white CCfe tube of 
this type operates at 6 or 8 watts per candle, the CQ& unfor- 
tunately giving much radiation of very low or totally negligible 
luminous value. Tubes filled with neon work at somewhat better 
than 1 watt per candle power but the gas is costly and trouble- 
some to work with. Thus, in spite of the improvements in 
illuminants during recent years, there is still much to be done 
in improving their efficiency, and especially in the smaller units. 
All the very high efficiencies yet attained have been with radiant 
sources of several hundred or even several thousand candle power. 

For everyday work the thing most needed is an efficient light 
of moderate candle power and moderate intrinsic brilliancy com- 
bined with low cost and good color. Save under special circum- 
stances, very powerful radiants are disadvantageous, particularly 
if of great intrinsic brilliancy. 

Casting about the field, it certainly appears at first glance as 
though most modern radiants had been developed in the wrong 
direction. In particular, electric lights have been steadily pushed 
in the direction of enormous working temperature and very great 
intrinsic brilliancy, gaming greatly in efficiency, of course, but 
losing in convenience. What is most wanted is not a light giving 
5000 candle power at 0.2 watt per candle, but one for ordinary 
voltages giving 5 or 10 candle power at even 1 watt per candle. 
The vacuum-tube lamp seems at present to give the greatest 
chance for revolutionary improvements, and even this seems to 
involve very serious difficulties. 

Similarly, in gaslights we have regenerative and mantle burners 
giving 50 or 100 candle power at a very good efficiency or press-gas 
burners of 1000 or more candle power at still very much higher effi- 
ciency, but they are too powerful and too bright to be entirely 
satisfactory, even were they open to no other objections. For most 
purposes, a Welsbach giving 15 candle power on 1 cubic foot of gas 
per hour would be vastly more useful than one giving 75 candle 
power on 4 cubic feet per hour. Of flame radiants, none save 
acetylene marks any material advance in recent years in point of 
easy applicability. It would seem that modern chemistry might 


achieve something of value in adding to the materials of illumina- 
tion. There is a group of substances possessing enormous power 
of giving off radiation when suitably stimulated. It is perhaps not 
too much to hope that some such material of extraordinary potency 
with respect to luminous rays may reward the pertinacious inves- 
tigator. There is no intrinsic reason why an exaggerated type of 
phosphorescence, capable of storing sunlight at a high efficiency, 
may not in due season be discovered. This would settle the arti- 
ficial lighting problem — unless the color were irremediably bad 
• — in a beautifully simple way. Or it might be possible to repro- 
duce by a commercial process the slow oxidation or analogous 
change responsible for the glowing of decaying wood and of 
certain microorganisms, and probably also for the light of the 
firefly and his allies. 

Whatever the method, the aim of improvement should be the 
production of efficient lights of moderate intensity and intrinsic 
brilliancy, coupled with good color, preferably capable of easy 

The steady tendency, as the art of illumination has advanced, 
has been towards more and more complete subdivision of the radi- 
ants, and the subordination of great brilliancy to perfect distribu- 
tion. One of the most important lessons of the Pan-American 
Exposition was Mr. Stieringer's demonstration of the magnificent 
usefulness of 8-c.p. incandescent lamps, skillfully installed. 

In the art of illumination, as much depends on the efficient use 
of lights as on the efficiency of the lights themselves. A tallow 
candle, just where it ought to be, is better than a misplaced arc 
lamp, and, even taking our present illuminants with all their limi- 
tations, skill will work wonders of economy. 

It is particularly in the direction of adroit use that the present 
path of progress lies. One of the fundamental facts in practical 
lighting, which has been many times suggested in these pages, and 
which lies at the root of improvements, is the need of keeping down 
intrinsic brilliancy. 

The true criterion of effective and efficient lighting is not simple 
illumination, which resolves itself into a pure matter of foot- 
candles, but visual usefulness, which takes account of the physio- 
logical factors in artificial lighting. 

If one denotes the illumination measured in foot-candles or other 
convenient units by /, then the visual usefulness is in part meas- 


ured by the product I<r, where <r is proportional to the effective area 
of the pupil. This of course is constantly shifting as the illumina- 
tion changes, but, broadly, it is an inverse function of the intrinsic 
brilliancy of the radiants used. Other physiological factors like 
adaptation also depend directly upon the intrinsic brilliancy to 
which the retina is exposed. The criterion thus becomes of the 

form i = 7-7m 9 where B is the intrinsic brilliancy of the radiant, 

and i is the visual usefulness, or the effective brilliancy of the 

Now as a matter of practice this is important, for it indicates 
that a badly placed arc light, for example, may actually work seri- 
ous injury to the effective illumination, and within reasonable 
limits one could fairly go as far as to say that the usefulness of an 
unmodified radiant varies inversely with its intrinsic brilliancy. 
Obviously, then, shading the radiant may gain useful illumi- 
nation, although it actually loses light, which in fact experience 
has shown to be the case. 

In electric lighting, incandescent lamps at 3 watts per candle, so 
disposed as to keep clear of the field of vision, are fully as valuable 
illuminants as lamps at 2 watts per candle wrongly installed, so as 
to either dazzle the eye or to require heavy shading to avoid it. 
Shaded they must be for hygienic reasons whenever visible. 

In actual practice it is a matter of great difficulty to place lights 
wholly out of the field of vision, and the more brilliant the lights are 
the greater necessity for shading them. Hence, it becomes a diffi- 
cult matter to treat modern illuminants without loss of efficiency. 

A very promising line of improvement in artificial lighting, and 
the one from which much may be expected in the near future, is 
indirect and semi-indirect lighting. As the intrinsic brilliancy 
of the source rises, the relative importance of diffusion increases, 
since shading, to be effective, must be denser. 

There is room for splendid developments in diffuse lighting, 
using arcs, Nernst lamps, incandescents of every sort, Welsbach 
mantles, and acetylene. In this way such radiants can be used 
with the full advantage of their great efficiency, and with good 
diffusion from white or nearly white surfaces the net efficiency can 
be fairly well maintained. As has already been noted, lighting 
by diffusion in ordinary interiors, where the surfaces are not gener- 
ally good, requires a very lavish use of light, but with a careful 


study of the conditions may come the possibility of efficient and 
beautiful lighting in which the radiants shall be effectively con- 

This method of working, too, has an artistic advantage, in that 
the light can be slightly modified by tinted diffusing surfaces with 
far greater success than by any arrangement of colored shades. 
The latter are not available in delicate and easily graduated 
shades, while pigments can be worked upon diffusing surfaces in 
almost any desired manner. 

The weak point of lighting by diffusion is the fact that the 
radiants are usually installed in rather inaccessible places, and the 
reflectors are certain to suffer from dust, unless special care is taken. 

It will be readily seen that the attainment of high luminous 
efficiency by means of driving illuminants to a very high specific 
brilliancy tends to defeat its own ends. If it costs, as it does, 
from 20 to 40 per cent of the luminous energy to secure diffusion 
complete enough to render the source suitable to use, then it is 
clear that it may be worth while to sacrifice a corresponding 
amount in luminous efficiency, in order to obtain a light of 
sufficiently low intrinsic brilliancy to be used without diffusion. 

Just how low intrinsic brilliancy is necessary to render the use 
of diffusers needless depends in no small measure on the amount of 
luminous energy which reaches the eye from the source considered. 
In other words, the physiological danger of glare from an illuminant 
is a function of the rate at which the retina has to take care of the 
energy which is delivered to it. Destructive and constructive work 
is continually being done at the retina, and the net result depends 
on the balance between these two factors. Neglect of this ques- 
tion of energy has led to a great deal of unnecessary alarm and 

As a matter of common experience, an arc light in a thin 
diffusing globe, of which the intrinsic brilliancy is conspicuously 
greater than could be tolerated at short range, is perfectly harm- 
less at the distance of a few hundred feet, while a source of con- 
siderably lower intrinsic brilliancy might be painful and harmful 
at close range. When the eye is in a state of full dark-adapta- 
tion, even very weak sources may produce harmful glare. The 
author has suffered from the misquotation of a paragraph in the 
first edition of this book, which set about 5 candle power per square 
inch as the highest permissible intrinsic brilliancy, although in 


the same paragraph he stated that half this value was preferable. 
Five candle power per square inch is perfectly safe out of doors 
or in large spaces, while even 2.5 may be excessive in lights at 
short range. In ordinary interior work it is preferable to keep 
the intrinsic brilliancies well below this figure, in extreme cases 
perhaps even below 1 candle power per square inch. 

Now, no unscreened illuminant, save the Moore tube, falls 
within this particular region, and it is to vacuum-tube or lumi- 
nescent lighting in one form or another that we chiefly must look 
for sources of intrinsic brilliancy low enough to permit them to be 
used unscreened. It seems doubtful at present whether they can 
be obtained at an efficiency which makes the game worth the candle. 
Considering the low intrinsic brilliancy of the Moore tube, how- 
ever, it compares more favorably with necessarily screened sources 
of light than its actual specific consumption in watts per candle 
would indicate. 

Aside from gaseous illuminants, the best chance for obtaining 
sources of low intrinsic brilliancy seems to be by chemical processes 
analogous to those carried on by photogenic bacteria and perhaps 
by the fireflies. Nothing practical has yet appeared in this par- 
ticular field. Certain luminescent phenomena akin to phosphor- 
escence have been the subject of some experiments, and are not 
without hope for useful results, although nothing substantial has 
yet been done. 

Broadly, then, future progress in efficient illumination depends 
either upon further increase in the luminous efficiency of intense 
sources, or, on the other hand, in the development of fairly 
efficient, less intense sources which make up by low intrinsic 
brilliancy for their losses in specific consumption. Improvements 
of the first sort have been rapid since the first edition of this 
book was published, and have now reached, as the figures given 
earlier indicate, a point where further progress is likely to be slow. 
The improvements in the near future are likely to be rather in 
length of life and steadiness of the luminous sources than in any 
conspicuous increase of efficiency. Along the second line of prog- 
ress there is perhaps a greater opportunity, albeit we do not know 
in what particular way it is likely to be brought to our notice. 

Meanwhile we must do the best we can, with the illuminants 
which are now at hand, to furnish light of suitable amount and 
quality. To sum up the suggestions repeatedly made in these 


pages, the commonest failings in present methods of lighting are a 
tendency to use too brilliant radiants and to make up in quantity 
what is lacking in quality. More study of the practical conditions 
of lighting and less blind faith in bright lights would generally 
both improve practical illumination and tend to economy. 

Imagine, for example, an attempt to light a billiard table where 
the balls had been stained to match the cloth. Yet this sort of 
thing, on a less aggravated scale, happens far oftener than would 
be thought possible. Even in buildings designed to fulfill hygienic 
conditions, sins against the fundamental principles of lighting are 
distressingly common. An observing writer has grimly designated 
modern schools " bad-eye factories," and certainly, even with the 
full advantage of natural light and buildings in which conditions 
ought to be favorable, the results are frequently bad. 

With artificial light the task of proper lighting is of increased 
difficulty, and is further complicated by the sometimes impossible 
requirements of the latest fashionable scheme of decoration. The 
best results can be attained only by constant attention to details 
and a keen perception of the conditions to be met. 

The illumination of the future ought to mean the intelligent 
use of the lights we now have, not less than the application of the 
lights which we may hope in the fullness of time to obtain. 



Abney's table of color differences, 28. 

Abolition of shadows, 19. 

Acetylene burner, 94. 

Acetylene gas, 91. 

Acetylene gas, cost of, 97. 

Adaptation of the eye, 6. 

Adoption of international standard 
candle, 58. 

After-images in the eye, 13. 

Air gas, 85. 

Air vitiation of various illuminants, 

Altar illumination, 256. 

Alternating- and direct-current arc 
lights, comparison of, 162. 

Alternating-current arc light, 159. 

Alt-market, Dresden, 288. 

Analysis of coal gas, 87. 

Apparatus for comparison of incan- 
descent lights, 67-69. 

Arc lamp, Blondel's flaming, 168. 

Arc light, alternating-current, 159. 

Arc light carbons, 153, 155. 

Arc light, comparison of direct- and 
alternating-current, 162. 

Arc light, current density and inten- 
sity, 152. 

Arc light, efficiency of, 158. 

Arc light, flaming, 164. 

Arc light, flaming, General Electric 
Company's, 170. 

Arc light, inclosed, distribution of 
light from, 157. 

Arc light, intensive, 163. 

Arc light, inverted, 274. 

Arc light, Jandus regenerating flame 
arc, 169. 

Arc light, luminous, 172. 

Arc light, magnetite, 172. 

Arc light, open, distribution of light 

from, 156. 
Arc lights, color of, 164. 
Arc lights, efficiencies of, 162. 
Arc lights in work shops, 247. 
Arc lights, open, 153. 
Arc lights, outdoor, 156. 
Arcs, magnetic, 312. 
Arcs, vertical carbon flame, 312. 
Argand gas burner, 89. 
Arrangement of interior lights, 213. 
Artificial light, early sources of, 77. 
Artificial lighting, fundamentals of, 

Auer light, 101. 


Basements, illumination of, 231. 

Basic facts in incandescent lamp prac- 
tice, 122. 

Bathrooms, illumination of, 231. 

Bedrooms, illumination of, 230. 

Berlin high pressure gas plant, 109. 

Billiard rooms, illumination of, 231. 

Blackboards, lighting of, 265. 

Blau-gas, 88. 

Blondel system lamps, 312. 

Blondel's flaming arc lamp, 168. 

Boston schoolroom illumination, 262. 

Bouguer's photometer, 60. 

Brackets, use of, in interior lighting, 

Bunsen burner, 103. 

Bunsen screen, 61. 

Bunsen photometer, 60-62. 

Burner, acetylene, 94. 

Burner, Bunsen, 103. 

Burner, oxyhydrogen, 100. 

Burning fluids of early days, 80. 





Candle, foot, definition of, 7. 

Candle, international standard, 57. 

Candle, meter, definition of, 7. 

Candle, parliamentary sperm, 53. 

Candles, illuminating, 81. 

Calcic carbide, 92. 

Carcel lamp, 53. 

Carbons, arc light, 153, 155. 

Ceiling lights, 217. 

Ceiling lights in halls, 238. 

Cellulose mantles, 110. 

Ceria, action of, in Welsbach man-* 

ties, 102. 
Ceria and color variation in mantles, 

Chandeliers, 223. 
Chandeliers, for churches, 253. 
Chevreul's experiments with tinted 

lights, 34. 
Church altar lighting, 256, 
Church chandeliers, 253. 
Churches, illumination of, 252. 
Classes of illuminants, 77. 
Clerical work, illumination for, 234. 
Closets, illumination of, 231. 
Coal gas, 86. 
Color absorption, 29. 
Color differences, Abney's table of, 

Color, fundamental law of, 25. 
Color of arc lights, 164. 
Color of incandescent electric lamps, 

Color of mantle burners, 112. 
Color of walls in practical illumina- 
tion, 51. 
Colored glass, luminosity of light 

through, 32. 
Colored illumination, 324. 
Colored lights, effects of, on colors, 29. 
Colored lights, general effects of, 33. 
Colors, effects of faint illumination on, 

Colors of common illuminants, 35. 
Colors of the solar spectrum, 26. 
Colors, variation of, under artificial 

light, 27. 

Colors viewed in colored lights, 29. 

Commercial candles, 82. 

Common troubles of mantle burn- 
ers, 114. 

Comparing incandescent lights, 67. 

Composition of petroleum, 80. 

Composition of Welsbach mantle, 101. 

Construction of the incandescent elec- 
tric lamp, 119. 

Construction of Nernst lamp glower, 

Consumption of gas in inverted burn- 
ers, 107. 

Consumption of gas in open and 
mantle burners, 105. 

Consumption of incandescent lamps, 

Contracts for street lighting, 314. 

Converging carbon lamps, 312. 

Cooper-Hewitt mercury vapor lamp, 

Copley Square, Boston, 288. 

Cost of manufacturing acetylene gas, 

Cost of various illuminants, 115. 

Cotton mantles, 110. 

Counting room illumination, 241. 

Current density and intensity in arc 
light, 152. 

Cut glass shades, 184. 

D'Arsonval acetylene gas generator, 

Davy introduces electric arc, 150. 
Daylight photometer, 21. 
Decorative circuits, temporary, 334. 
Decorative fixtures, 224. 
Decorative illumination, 316-335. 
Delivery rooms, illumination of, 269. 
Determining amount of illumination 

necessary, 4. 
Development of the incandescent 

lamp, 100. 
Diffuse reflection, 38. 
Diffuse reflection from colored papers, 




Diffusion in interior illumination, 210. 

Dining rooms, illumination of, 228. 

Direct- and alternating-current arc 
lights, comparison of, 162. 

Direct-indirect reflectors, 206. 

Direct vs. indirect system for offices, 

Disk, the Leeson, 62. 

Distribution of artificial light affect- 
ing the eye, 15. 

Distribution of interior lights, 215. 

Distribution of light from an open 
arc, 156. 

Distribution of light from inclosed 
arc light, 157. 

Distribution of street light, 302. 

Dressing tables, illumination of, 230. 

Drummond light, 99. 

Domes, illumination of, 251. 

Domestic lighting, 207-232. 

Domestic lighting, important rule for, 


Earliest sources of artificial light, 77. 

Economics of the incandescent lamp, 

Economy in street lighting, 298. 

Efficiencies of arc lights, 162. 

Efficiencies of utilization, 245. 

Efficiency, 338. 

Efficiency and temperature in incan- 
descent lamps, 128. 

Efficiency in incandescent electric 
lamps, 130. 

Efficiency of commercial incandes- 
cent lamps, 126. 

Efficiency of electric arc light, 158. 

Electric arc light, principle of, 150. 

Elliot lamp, 59. 

English schoolroom lighting, 265. 

Exposition buildings, 277. 

Extensive reflectors, 199. 

Exterior illumination, 279-315. 

Eye, human, and light, 2, 5, 6, 13. 

Eye, human, iris diaphragm, 13. 

Eye, human, variation of pupil, 14. 

Fabrics, reflection from, 46-48. 
Facade illumination, 320. 
Factors in interior illumination, 209. 
Faint illumination, effect of, on colors, 

Fechner'8 fraction, 4. 
Fechner's law, 3. 
Filaments, first attempts, 116. 
Filaments, forms of, 120, 121. 
Filaments, looped, 124. 
Filaments, manufacture of, 117. 
Filaments, metallized, 136. 
Filaments, osmium, 136. 
Filaments, tantalum, 137. 
Filaments, tungsten, 138. 
First-class streets, lighting, 296. 
First public street lighting, 291. 
Fixtures, decorative, 224. 
Flame illuminants, 79. 
Flaming arc light, 164. 
Flat-flame gas burners, 89. 
Flicker photometer, 64. 
Flickering lights, 16, 17. 
Flux in street lighting, 305. 
Flux of light method of computation, 

Flux, luminous, unit of, 10. 
Foot-candle, definition of, 7. 
Fraction, Fechner's, 4. 
Fraunhofer lines, 26. 
Frieze illumination, 250. 
Frieze lights in halls, 238. 
Fundamental law of color. 25. 
Fundamentals of artificial lighting, 11. 

Gas, acetylene, 91. 
Gas, acetylene, cost of, 97. 
Gas, air, 85. 

Gas burner, Argand, 89. 
Gas burner, flat-flame, 89. 
Gas burner, Siemens regenerative, 90. 
Gas burner, Wenham, 90. 
Gas burners, 88. 
Gas, coal, 86. 

Gas consumption in open and mantle 
burners, 105. 



Gas lights, high pressure, 107. 

Gas lights in shops, 248. 

Gas machines, 85. 

Gas, Pintsch, 88. 

Gas, water, 87. 

Gasoline gas machine, 85. 

General Electric Company's flame- 
arc lamp, 170. 

General illumination and reflection, 

Generators, acetylene, 94-95. 

Glass, colored luminosity of light 
transmitted through, 32. 

Glass shades, 187. 

Globe, holophane, 191. 

Globes, light absorption of various 
kinds, 186. 

Goggles, Indian, 2. 

Grouping lights in illumination of 
halls, 237. 


Hallways, illumination of, 225. 
Halls, illumination of, 236. 
Harcourt pentane standard, 54. 
Hefner lamp, 53, 54. 
Height of street lights, 303. 
Heterochromic photometry, 66. 
Hewitt's fluorescent reflecting screen, 

High pressure gas lights, 107. 
High room illumination, 219, 249. 
Holophane globes, 191. 
Houston and Kennclly's illuminome- 

ter, 74. 
Human eye, the, 2, 5, 6, 13. 
Hygienic relations of illuminants, 



Illuminants, acetylene gas, 91-94. 
Illuminants, common, colors of, 35. 
Illuminants, comparative cost of, 115. 
Illuminants, composition of, 77. 
Illuminants, flame, 79. 
Illuminants, hygienic relations of, 

Illuminants, interior, choice of, 215. 
Illuminants, petroleum, 80. 
Illuminants, street, modern varieties, 

Illuminating gases, 87. 
Illuminating system in Boston school- 
rooms, 263. 
Illumination, artificial, key to, 3. 
Illumination for high rooms, 219. 
Illumination for machines, 244. 
Illumination for public buildings, 258. 
Illumination for work rooms, 243. 
Illumination, indirect, 203. 
Illumination of basements, 231. 
Illumination of bathrooms, 231. 
Illumination of bedrooms, 230. 
Illumination of billiard rooms, 231. 
Illumination of churches, 252. 
Illumination of closets, 231. 
Illumination of dining rooms, 228. 
Illumination of domes, 251. 
Illumination of halls, 236. 
Illumination of hallways, 225. 
Illumination of kitchens, 229. 
Illumination of large rooms, 235-278. 
Illumination of libraries, 227. 
Illumination of library buildings, 267. 
Illumination of living-rooms, 228. 
Illumination of music rooms, 226. 
Illumination of offices, 233. 
Illumination of pantries, 229. 
Illumination of public rooms, 269. 
Illumination of public squares, 281. 
Illumination of reception rooms, 226. 
Illumination of schoolhouses, 261. 
Illumination of shops, 247. 
Illumination of tennis courts, 265. 
Illumination of the Mosque of St. 

Sophia, 255. 
Illumination of theaters, 259. 
Illumination, strength of, in relation, 

to shade perception, 5. 
Illumination, strength of, required for, 

various needs, 20. 
Illumination, to determine amount 

necessary, 4. 
Illumination, two general purposes 

of, 1. 



Illuminometer, Houston and Ken- 
nelly's, 74. 

Incandescent electric illumination, 
basic facts, 122. 

Incandescent illuminante, 99. 

Incandescent electric lamps, 116-149. 

Incandescent electric lamps, color of, 

Incandescent electric lamps, con- 
sumption of, 136. 

Incandescent electric lamps, effi- 
ciency of, 126, 130. 

Incandescent electric lamps, sizes of, 

Incandescent electric lamp, econom- 
ics of, 132. 

Incandescent electric lamps, meas- 
uring, 123. 

Incandescent lamps, photometering, 

Inclosed arc light, principle of, 153. 

Indian goggles, 2. 

Indirect illumination, 203. 

Indirect illumination in large inte- 
riors, 275. 

Indirect lighting for offices, 242. 

Indirect vs. direct system for offices, 

Intensive arc light, 163. 

Intensive reflectors, 199. 

Interior decorating, 331. 

Interior illuminante, choice of, 215. 

Interior illumination, factors in, 209. 

Interior illumination, diffusion in, 

Interior lights, arrangement of, 213. 

Interior lights, distribution of, 215. 

International standard candle, 57. 

Intrinsic brightness, definition of, 11. 

Inverted arc light, 273. 

Inverted mantle burners, 105. 

Inverted reflectors, 205. 

Iris, action of, in various lights, 15. 

Iris diaphragm of the eye, 13. 

Jandus regenerating flame lamp, 169. 
Junior Welsbach light, 104. 


Kerosene lamps, 83. 

Key to artificial illumination, 3. 

Kitchens, illumination of, 229. 

Lamp, Blondel flaming arc, 168. 
Lamp, Carcel, 53. 
Lamp, Cooper-Hewitt mercury vapor, 

Lamp, Elliot, 59. 
Lamp, flaming arc, mechanism of, 

Lamp, General Electric Company's 

flame-arc, 170. 
Lamp, Hefner, 53, 54. 
Lamp, incandescent, development of, 

Lamp, Jandus regenerating flame 

arc, 169. 
Lamp, magnetic, 313. 
Lamp, magnetite arc, 172. 
Lamp, Nernst, 144. 
Lamps, osmium, 136. 
Lamp, quartz-mercury, 179. 
Lamp, tantalum, 137. 
Lamp, titanium-carbide arc, 176. 
Lamps, tungsten, 140. 
Lamps, converging carbon, 312. 
Lamps, earliest patterns, 78. 
Lamps for outlining, 330. 
Lamps, incandescent, photometering, 

Lamps, magnetic arc, 312. 
Lamps, miniature, 332. 
Lamps, oil, 82. 
Lamps, Rochester, 83. 
Lamps, street, location of, 308. 
Lamps, system, Blondel, 312. 
Large rooms, illumination of, 235. 
Law, Fechner's, 3. 
Law, fundamental, of color, 25. 
Law of inverse squares, 8. 
Law of regular reflection, 37. 
Leeson disk, 62. 
Libraries, illumination of, 227. 
Library buildings, illumination of, 




Library stacks, lighting, 269. 

Life of mantles, 111. 

Light absorption of various globes, 

Light and the eye, 10-24. 

Light, arc, alternating-current, 159. 

Light, artificial, early sources of, 77. 

Light, flaming arc, 164. 

Light, inclosed arc, distribution of 
light from, 157. 

Light, intensive arc, 163. 

Light, luminosity of, 32. 

Light, luminous arc, 172. 

Light, magnetite arc, 172. 

Light measurement, questions in- 
volved, 52. 

Light, open arc, distribution of light 
from, 156. 

Light, quartz-mercury, 180. 

Light, the Moore tube, 181. 

Light, Welsbach, 101. 

Lighting by inverted arcs, 273. 

Lighting, direct-indirect, 206. 

Lighting, domestic, important rule 
for, 220. 

Lighting, domestic, 207-232. 

Lighting high rooms, 249. 

Lighting, indirect, 205. 

Lighting schoolrooms, 263. 

Lighting, spotted, remedy for, 243. 

Lighting tennis courts, 265. 

Lights, arc, 150-183. 

Lights, ceiling, 217. 

Lights for signals, 32. 

Lights, interior, arrangement of, 213. 

Lights, open arc, 153. 

Lights, side-wail, 218. 

Lime lights, 99. 

Living rooms, illumination of, 228. 

Location of ceiling lights, 217. 

Location of street lamps, 308. 

Looped filaments, 124. 

Lucigen torch, 84. 

Lumen, definition of, 10. 

Luminous arc light, 172. 

Luminous flux, unit of, 10. 

Lummer-Brodhun photometer, 60, 63. 

Lux, as a unit of illumination, 10. 


McCreary shade, 189. 

Machine illumination, 244. 

Magnetic arc lamps, 312. 

Magnetite arc light, 172. 

Malignani process of exhausting 
bulbs, 120. 

Mantle burners, 103-106. 

Mantle burners, color of, 112. 

Mantle manufacture, materials used, 

Mantles, color variation in, 113. 

Mantles, common troubles, 114. 

Mantles, cotton, 110. 

Mantles, life of, 111. 

Mantles, silk, 110. 

Mantles, Welsbach, composition of, 

Manufacture of filaments, 117, 121. 

Manufacturing cost of acetylene gas, 

Material in mantle manufacture, 110. 

Matthew's integrating photometer, 

Measurement of light, questions in- 
volved, 52. 

Measuring daylight, 21. 

Measuring incandescent lamps, 123. 

Measuring intrinsic brightness, 12. 

Measuring street illumination, 310. 

Mechanism of flaming arc lamps, 

Mercury vapor lamp, 177. 

Metallic oxide lamps, 313. 

Metallized filaments, 136. 

Meter-candle, definition of, 7. 

Methods of determining unit of illu- 
mination, 10. 

Methods of interior illumination, 208. 

Methods of lighting halls, 238-240. 

Methven screen, 59. 

Miniature incandescent lamps, 332. 

Moonlight schedules, 299. 

Moore tube, 181, 339. 

Mosque of St. Sophia, lighting of, 

Multiple-looped filaments, 126. 

Multiple reflection, 43. 




Museums, lighting, 271. 

Music rooms, illumination of, 226. 

Nernst lamp, 144. 

Offices, illumination of, 233. 
Oil lamps, 82. 
Osmium filaments, 136. 
Outdoor arc lights, 156. 
Outlets for hall illumination, 240. 
Outlets in interior illumination, 222. 
Outlining illumination, 325. 
Oxides in magnetite arc lamps, 176. 


Pantries, illumination of, 229. 
Paper, colored, reflective qualities, 49. 
Park lighting, 280, 289. 
Parliamentary sperm candle, 53. 
Pentane standard, 54. 
Periodical rooms, illumination of, 270. 
Petroleum as an illuminant, 80. 
Petroleum products, 81. 
Photometer, daylight, 21. 
Photometer, Matthews' integrating, 

Photometer, portable, 71-74. 
Photometers, the Bouguer, 60. 
Photometer, the Bunsen, 60-62. 
Photometer, the "flicker," 64. 
Photometer, the Lummer-Brodhun, 

Photometer, the reading, 75. 
Photometer, the Simmance-Abady, 

Photometering incandescent lamps, 

Photometrical standards, 53. 
Photometry, 59. 
Photometry, heterochromic, 66. 
Pintsch gas, 88. 
Place de la Concorde, illumination of, 

Platinum in electric lamp, 119. 
Point-to-point method, 70. 
Portable lights, acetylene, 96. 

Portable lights, Lucigen torch, 84. 
Portable photometer, 71-74. 
Portable photometer, Weber's, 70-72. 
Potsdamer Platz, Berlin, 282. 
Press-gas lighting, 109. 
Principle of the arc light, 150. 
Principle of the inclosed arc light, 153. 
Prismatic reflectors, 196. 
Products of petroleum, 81. 
Public building illumination, 258. 
Public rooms, illumination of, 269. 
Public squares, 280. 
Pupil of the eye, variation of, 14. 


Quartz-mercury lamp, 179. 
Quai de Mt. Blanc, 317. 


Radiants for street lighting, 303. 

Railway stations, 272. 

Ramie fiber mantles, 110. 

Rare earths in Welsbach mantles, 101. 

Reading lamps, 223. 

Reading photometer, 75. 

Receptacles for decorative designs, 

Reception rooms, illumination of, 226. 

Reflected spectra, 26 

Reflecting cove, 203. 

Reflecting shades, 188. 

Reflecting surfaces, table of results 
from, 45. 

Reflection, diffuse, 38. 

Reflection, diffuse, from various sur- 
faces, 49, 50. 

Reflection from fabrics, 46-48. 

Reflection from colored papers, 49. 

Reflection in general illumination, 

Reflection, multiple, 43. 

Reflection, regular or specular, 37. 

Reflection, selective, 36, 42. 

Reflection, Trotter's experiments in, 

Reflectors, cove, 203. 

Reflectors, extensive, 199. 

Reflectors, intensive, 199. 



Reflectors, inverted, 205. 
Reflectors, prismatic, 196. 
Reflectors, X-ray, 205. 
Regenerative gas burners, 90. 
Remedy for spotted lighting, 243. 
Rochester lamp, 83. 
Roman lamps, 78. 
Rule for domestic lighting, 220. 


Scenic illumination, 310-335. 

Schoolhouses, lighting of, 261. 

Schoolrooms in England, method of 
lighting, 265. 

Schoolrooms in Boston, method of 
lighting, 264. 

Screen, Hewitt's fluorescent reflect- 
ing, 179. 

Screening street lamps, 310. 

Search lights for surface lighting, 320. 

Second-class streets, lighting, 297. 

Selection of interior illuminants, 215. 

Selective reflection, 36, 42. 

Series magnetite lamp, 174. 

Shades, objectionable, 185. 

Shade perception, 5. 

Shade, the McCreary, 189. 

Shades and reflectors, 184-206. 

Shadows, abolition of, 19. 

Shops, illumination of, 247. 

Shotgun diagrams, 133. 

Side lights, 218. 

Siemens regenerative burner, 90. 

Signals, lights for, 32 

Silk mantles, 110. 

Simmance-Abady photometer, 65. 

Sizes of incandescent electric lamnfl. 

Snow blindness, 2. 

Solar spectrum, 26. 

Sources, earliest, of artificial light, 77. 

Spacing of street lights, 303. 

Special reflectors, 203. 

Spectra, reflected, 27. 

Specular reflection, 37. 

Spot light for surface illumination, 

Spotted lighting, remedy for, 243. 

Standard, the pentane, 54. 

Standard, Violle's platinum, 57. 

Street decorating, 328. 

Street illuminants, 312. 

Street illumination, measuring, 310. 

Street lamps, height of, 307. 

Street lamp, screening, 310. 

Street lighting, 290. 

Street lighting, contracts for, 314. 

Street lighting, distribution of, 302. 

Strength of illumination required for 

various work, 20. 
Structural illumination, 318. 
Structural limitations in interior 

lighting, 216. 
Surface illumination, 319. 
Switches for interior lighting, 232. 
System Blondel lamps, 312. 


Table, Abney's, of color differences, 

Table, Chevreul's experiments with 
tinted lights, 34. 

Table, comparison of direct- and al- 
ternating-current arcs, 162. 

Table of intrinsic brilliancies in can- 
dle power per square inch, 12. 

Table of Nernst lamp data, 148. 

Table, petroleum products, 81. 

Table, relations between various pri- 
mary standards, 58. 

Table, results from various reflecting 
surfaces, 45. 

Tantalum lamp, 137. 

Temperature and incandescent lamp 
efficiency, 128. 

Temporary decorative circuits, 334. 

Temporary illumination, 272. 

Tennis courts, illumination of, 265. 

Thames Embankment, 317. 

Theater illumination, 259. 

Theory of the Ulbricht sphere, 70. 

Third-class streets, lighting, 297. 

Titanium-carbide arc lamp, 176. 

Torch, Lucigen, 84. 

Trafalgar square, illumination of, 281. 

Trotter's experiments in reflection, 41. 



Tungsten filament, 138. 
Tungsten lamps, 140. 
Tungsten lamps in parks, 290. 
Tungsten lamp in domestic lighting, 

Tungsten street lamp, 314. 

Ulbricht sphere, 70. 
Unilateral illumination, 275. 
Units of illumination, 10. 
Utilization, efficiencies of, 245. 

Vacuum tube lamps, 336. 
Variation in commercial incandescent 

lamps, 134, 135. 
Variation of color in mantles, 113. 
Variation of colore under artificial 

light, 27. 
Variation of "law of inverse squares, 1 ' 

Variation of the pupil of the eye, 14. 

Vertical carbon flame arcs, 312. 
Violle's platinum standard, 57. 
Visual usefulness, 340. 
Vitiation of air by various Olumi- 
nants, 115. 


Water gas, 87. 

Weber's portable photometer, 70-72. 

Weber's tests of incandescent lamp 

efficiency, 127. 
Welsbach, Junior, light, 104. 
Welsbach light, 101. 
Wenham gas burner, 90. 
Work-room illumination, 243. 
Work-shop illumination, 247. 


X-ray reflectors, 205. 

Yellow components in light, 32.